|
|
|
|
#define _POSIX_C_SOURCE 200809L
|
|
|
|
|
#include <assert.h>
|
|
|
|
|
#include <errno.h>
|
|
|
|
|
#ifdef __linux__
|
|
|
|
|
#include <linux/input-event-codes.h>
|
|
|
|
|
#elif __FreeBSD__
|
|
|
|
|
#include <dev/evdev/input-event-codes.h>
|
|
|
|
|
#endif
|
|
|
|
|
#include <strings.h>
|
|
|
|
|
#include <time.h>
|
|
|
|
|
#include <wlr/types/wlr_cursor.h>
|
|
|
|
|
#include <wlr/types/wlr_output_layout.h>
|
|
|
|
|
#include <wlr/types/wlr_xcursor_manager.h>
|
|
|
|
|
#include "log.h"
|
|
|
|
|
#include "config.h"
|
|
|
|
|
#include "sway/debug.h"
|
|
|
|
|
#include "sway/desktop.h"
|
|
|
|
|
#include "sway/input/cursor.h"
|
|
|
|
|
#include "sway/input/input-manager.h"
|
|
|
|
|
#include "sway/input/keyboard.h"
|
|
|
|
|
#include "sway/input/seat.h"
|
|
|
|
|
#include "sway/ipc-server.h"
|
|
|
|
|
#include "sway/layers.h"
|
|
|
|
|
#include "sway/output.h"
|
|
|
|
|
#include "sway/tree/arrange.h"
|
|
|
|
|
#include "sway/tree/container.h"
|
|
|
|
|
#include "sway/tree/root.h"
|
|
|
|
|
#include "sway/tree/view.h"
|
|
|
|
|
#include "sway/tree/workspace.h"
|
|
|
|
|
|
|
|
|
|
static void seat_device_destroy(struct sway_seat_device *seat_device) {
|
|
|
|
|
if (!seat_device) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
sway_keyboard_destroy(seat_device->keyboard);
|
|
|
|
|
wlr_cursor_detach_input_device(seat_device->sway_seat->cursor->cursor,
|
|
|
|
|
seat_device->input_device->wlr_device);
|
|
|
|
|
wl_list_remove(&seat_device->link);
|
|
|
|
|
free(seat_device);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_destroy(struct sway_seat *seat) {
|
|
|
|
|
struct sway_seat_device *seat_device, *next;
|
|
|
|
|
wl_list_for_each_safe(seat_device, next, &seat->devices, link) {
|
|
|
|
|
seat_device_destroy(seat_device);
|
|
|
|
|
}
|
|
|
|
|
sway_cursor_destroy(seat->cursor);
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
wl_list_remove(&seat->new_node.link);
|
|
|
|
|
wl_list_remove(&seat->new_drag_icon.link);
|
|
|
|
|
wl_list_remove(&seat->link);
|
|
|
|
|
wlr_seat_destroy(seat->wlr_seat);
|
|
|
|
|
free(seat->prev_workspace_name);
|
|
|
|
|
free(seat);
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void seat_node_destroy(struct sway_seat_node *seat_node) {
|
|
|
|
|
wl_list_remove(&seat_node->destroy.link);
|
|
|
|
|
wl_list_remove(&seat_node->link);
|
|
|
|
|
free(seat_node);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Activate all views within this container recursively.
|
|
|
|
|
*/
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void seat_send_activate(struct sway_node *node, struct sway_seat *seat) {
|
|
|
|
|
if (node_is_view(node)) {
|
|
|
|
|
if (!seat_is_input_allowed(seat, node->sway_container->view->surface)) {
|
|
|
|
|
wlr_log(WLR_DEBUG, "Refusing to set focus, input is inhibited");
|
|
|
|
|
return;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
view_set_activated(node->sway_container->view, true);
|
|
|
|
|
} else {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
list_t *children = node_get_children(node);
|
|
|
|
|
for (int i = 0; i < children->length; ++i) {
|
|
|
|
|
struct sway_container *child = children->items[i];
|
|
|
|
|
seat_send_activate(&child->node, seat);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* If con is a view, set it as active and enable keyboard input.
|
|
|
|
|
* If con is a container, set all child views as active and don't enable
|
|
|
|
|
* keyboard input on any.
|
|
|
|
|
*/
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void seat_send_focus(struct sway_node *node, struct sway_seat *seat) {
|
|
|
|
|
seat_send_activate(node, seat);
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_view *view = node->type == N_CONTAINER ?
|
|
|
|
|
node->sway_container->view : NULL;
|
|
|
|
|
|
|
|
|
|
if (view && seat_is_input_allowed(seat, view->surface)) {
|
|
|
|
|
#if HAVE_XWAYLAND
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (view->type == SWAY_VIEW_XWAYLAND) {
|
|
|
|
|
struct wlr_xwayland *xwayland = server.xwayland.wlr_xwayland;
|
|
|
|
|
wlr_xwayland_set_seat(xwayland, seat->wlr_seat);
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
struct wlr_keyboard *keyboard = wlr_seat_get_keyboard(seat->wlr_seat);
|
|
|
|
|
if (keyboard) {
|
|
|
|
|
wlr_seat_keyboard_notify_enter(seat->wlr_seat,
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
view->surface, keyboard->keycodes,
|
|
|
|
|
keyboard->num_keycodes, &keyboard->modifiers);
|
|
|
|
|
} else {
|
|
|
|
|
wlr_seat_keyboard_notify_enter(
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
seat->wlr_seat, view->surface, NULL, 0, NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
void seat_for_each_node(struct sway_seat *seat,
|
|
|
|
|
void (*f)(struct sway_node *node, void *data), void *data) {
|
|
|
|
|
struct sway_seat_node *current = NULL;
|
|
|
|
|
wl_list_for_each(current, &seat->focus_stack, link) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
f(current->node, data);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_container *seat_get_focus_inactive_view(struct sway_seat *seat,
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *ancestor) {
|
|
|
|
|
if (ancestor->type == N_CONTAINER && ancestor->sway_container->view) {
|
|
|
|
|
return ancestor->sway_container;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *current;
|
|
|
|
|
wl_list_for_each(current, &seat->focus_stack, link) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *node = current->node;
|
|
|
|
|
if (node->type == N_CONTAINER && node->sway_container->view &&
|
|
|
|
|
node_has_ancestor(node, ancestor)) {
|
|
|
|
|
return node->sway_container;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void handle_seat_node_destroy(struct wl_listener *listener, void *data) {
|
|
|
|
|
struct sway_seat_node *seat_node =
|
|
|
|
|
wl_container_of(listener, seat_node, destroy);
|
|
|
|
|
struct sway_seat *seat = seat_node->seat;
|
|
|
|
|
struct sway_node *node = seat_node->node;
|
|
|
|
|
struct sway_node *parent = node_get_parent(node);
|
|
|
|
|
struct sway_node *focus = seat_get_focus(seat);
|
|
|
|
|
|
|
|
|
|
if (node->type == N_WORKSPACE) {
|
|
|
|
|
seat_node_destroy(seat_node);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Even though the container being destroyed might be nowhere near the
|
|
|
|
|
// focused container, we still need to set focus_inactive on a sibling of
|
|
|
|
|
// the container being destroyed.
|
|
|
|
|
bool needs_new_focus = focus &&
|
|
|
|
|
(focus == node || node_has_ancestor(focus, node));
|
|
|
|
|
|
|
|
|
|
seat_node_destroy(seat_node);
|
|
|
|
|
|
|
|
|
|
if (!parent) {
|
|
|
|
|
// Destroying a container that is no longer in the tree
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Find new focus_inactive (ie. sibling, or workspace if no siblings left)
|
|
|
|
|
struct sway_node *next_focus = NULL;
|
|
|
|
|
while (next_focus == NULL) {
|
|
|
|
|
struct sway_container *con =
|
|
|
|
|
seat_get_focus_inactive_view(seat, parent);
|
|
|
|
|
next_focus = con ? &con->node : NULL;
|
|
|
|
|
|
|
|
|
|
if (next_focus == NULL && parent->type == N_WORKSPACE) {
|
|
|
|
|
next_focus = parent;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
parent = node_get_parent(parent);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (next_focus->type == N_WORKSPACE &&
|
|
|
|
|
!workspace_is_visible(next_focus->sway_workspace)) {
|
|
|
|
|
// Do not change focus to a non-visible workspace
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (needs_new_focus) {
|
|
|
|
|
// The structure change might have caused it to move up to the top of
|
|
|
|
|
// the focus stack without sending focus notifications to the view
|
|
|
|
|
seat_send_focus(next_focus, seat);
|
|
|
|
|
seat_set_focus(seat, next_focus);
|
|
|
|
|
} else {
|
|
|
|
|
// Setting focus_inactive
|
|
|
|
|
focus = seat_get_focus_inactive(seat, &root->node);
|
|
|
|
|
seat_set_raw_focus(seat, next_focus);
|
|
|
|
|
if (focus->type == N_CONTAINER) {
|
|
|
|
|
seat_set_raw_focus(seat, &focus->sway_container->workspace->node);
|
|
|
|
|
}
|
|
|
|
|
seat_set_raw_focus(seat, focus);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static struct sway_seat_node *seat_node_from_node(
|
|
|
|
|
struct sway_seat *seat, struct sway_node *node) {
|
|
|
|
|
if (node->type == N_ROOT || node->type == N_OUTPUT) {
|
|
|
|
|
// these don't get seat nodes ever
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *seat_node = NULL;
|
|
|
|
|
wl_list_for_each(seat_node, &seat->focus_stack, link) {
|
|
|
|
|
if (seat_node->node == node) {
|
|
|
|
|
return seat_node;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
seat_node = calloc(1, sizeof(struct sway_seat_node));
|
|
|
|
|
if (seat_node == NULL) {
|
|
|
|
|
wlr_log(WLR_ERROR, "could not allocate seat node");
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
seat_node->node = node;
|
|
|
|
|
seat_node->seat = seat;
|
|
|
|
|
wl_list_insert(seat->focus_stack.prev, &seat_node->link);
|
|
|
|
|
wl_signal_add(&node->events.destroy, &seat_node->destroy);
|
|
|
|
|
seat_node->destroy.notify = handle_seat_node_destroy;
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
return seat_node;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void handle_new_node(struct wl_listener *listener, void *data) {
|
|
|
|
|
struct sway_seat *seat = wl_container_of(listener, seat, new_node);
|
|
|
|
|
struct sway_node *node = data;
|
|
|
|
|
seat_node_from_node(seat, node);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void drag_icon_damage_whole(struct sway_drag_icon *icon) {
|
|
|
|
|
if (!icon->wlr_drag_icon->mapped) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
desktop_damage_surface(icon->wlr_drag_icon->surface, icon->x, icon->y, true);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void drag_icon_update_position(struct sway_drag_icon *icon) {
|
|
|
|
|
drag_icon_damage_whole(icon);
|
|
|
|
|
|
|
|
|
|
struct wlr_drag_icon *wlr_icon = icon->wlr_drag_icon;
|
|
|
|
|
struct sway_seat *seat = icon->seat;
|
|
|
|
|
struct wlr_cursor *cursor = seat->cursor->cursor;
|
|
|
|
|
if (wlr_icon->is_pointer) {
|
|
|
|
|
icon->x = cursor->x;
|
|
|
|
|
icon->y = cursor->y;
|
|
|
|
|
} else {
|
|
|
|
|
struct wlr_touch_point *point =
|
|
|
|
|
wlr_seat_touch_get_point(seat->wlr_seat, wlr_icon->touch_id);
|
|
|
|
|
if (point == NULL) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
icon->x = seat->touch_x;
|
|
|
|
|
icon->y = seat->touch_y;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
drag_icon_damage_whole(icon);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void drag_icon_handle_surface_commit(struct wl_listener *listener,
|
|
|
|
|
void *data) {
|
|
|
|
|
struct sway_drag_icon *icon =
|
|
|
|
|
wl_container_of(listener, icon, surface_commit);
|
|
|
|
|
drag_icon_update_position(icon);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void drag_icon_handle_map(struct wl_listener *listener, void *data) {
|
|
|
|
|
struct sway_drag_icon *icon = wl_container_of(listener, icon, map);
|
|
|
|
|
drag_icon_damage_whole(icon);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void drag_icon_handle_unmap(struct wl_listener *listener, void *data) {
|
|
|
|
|
struct sway_drag_icon *icon = wl_container_of(listener, icon, unmap);
|
|
|
|
|
drag_icon_damage_whole(icon);
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void drag_icon_handle_destroy(struct wl_listener *listener, void *data) {
|
|
|
|
|
struct sway_drag_icon *icon = wl_container_of(listener, icon, destroy);
|
|
|
|
|
icon->wlr_drag_icon->data = NULL;
|
|
|
|
|
wl_list_remove(&icon->link);
|
|
|
|
|
wl_list_remove(&icon->surface_commit.link);
|
|
|
|
|
wl_list_remove(&icon->unmap.link);
|
|
|
|
|
wl_list_remove(&icon->destroy.link);
|
|
|
|
|
free(icon);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void handle_new_drag_icon(struct wl_listener *listener, void *data) {
|
|
|
|
|
struct sway_seat *seat = wl_container_of(listener, seat, new_drag_icon);
|
|
|
|
|
struct wlr_drag_icon *wlr_drag_icon = data;
|
|
|
|
|
|
|
|
|
|
struct sway_drag_icon *icon = calloc(1, sizeof(struct sway_drag_icon));
|
|
|
|
|
if (icon == NULL) {
|
|
|
|
|
wlr_log(WLR_ERROR, "Allocation failed");
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
icon->seat = seat;
|
|
|
|
|
icon->wlr_drag_icon = wlr_drag_icon;
|
|
|
|
|
wlr_drag_icon->data = icon;
|
|
|
|
|
|
|
|
|
|
icon->surface_commit.notify = drag_icon_handle_surface_commit;
|
|
|
|
|
wl_signal_add(&wlr_drag_icon->surface->events.commit, &icon->surface_commit);
|
|
|
|
|
icon->unmap.notify = drag_icon_handle_unmap;
|
|
|
|
|
wl_signal_add(&wlr_drag_icon->events.unmap, &icon->unmap);
|
|
|
|
|
icon->map.notify = drag_icon_handle_map;
|
|
|
|
|
wl_signal_add(&wlr_drag_icon->events.map, &icon->map);
|
|
|
|
|
icon->destroy.notify = drag_icon_handle_destroy;
|
|
|
|
|
wl_signal_add(&wlr_drag_icon->events.destroy, &icon->destroy);
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
wl_list_insert(&root->drag_icons, &icon->link);
|
|
|
|
|
|
|
|
|
|
drag_icon_update_position(icon);
|
|
|
|
|
seat_end_mouse_operation(seat);
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void collect_focus_iter(struct sway_node *node, void *data) {
|
|
|
|
|
struct sway_seat *seat = data;
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *seat_node = seat_node_from_node(seat, node);
|
|
|
|
|
if (!seat_node) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
wl_list_remove(&seat_node->link);
|
|
|
|
|
wl_list_insert(&seat->focus_stack, &seat_node->link);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void collect_focus_workspace_iter(struct sway_workspace *workspace,
|
|
|
|
|
void *data) {
|
|
|
|
|
collect_focus_iter(&workspace->node, data);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void collect_focus_container_iter(struct sway_container *container,
|
|
|
|
|
void *data) {
|
|
|
|
|
collect_focus_iter(&container->node, data);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_seat *seat_create(const char *seat_name) {
|
|
|
|
|
struct sway_seat *seat = calloc(1, sizeof(struct sway_seat));
|
|
|
|
|
if (!seat) {
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
seat->wlr_seat = wlr_seat_create(server.wl_display, seat_name);
|
|
|
|
|
if (!sway_assert(seat->wlr_seat, "could not allocate seat")) {
|
|
|
|
|
free(seat);
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
seat->wlr_seat->data = seat;
|
|
|
|
|
|
|
|
|
|
seat->cursor = sway_cursor_create(seat);
|
|
|
|
|
if (!seat->cursor) {
|
|
|
|
|
wlr_seat_destroy(seat->wlr_seat);
|
|
|
|
|
free(seat);
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// init the focus stack
|
|
|
|
|
wl_list_init(&seat->focus_stack);
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
root_for_each_workspace(collect_focus_workspace_iter, seat);
|
|
|
|
|
root_for_each_container(collect_focus_container_iter, seat);
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
wl_signal_add(&root->events.new_node, &seat->new_node);
|
|
|
|
|
seat->new_node.notify = handle_new_node;
|
|
|
|
|
|
|
|
|
|
wl_signal_add(&seat->wlr_seat->events.new_drag_icon, &seat->new_drag_icon);
|
|
|
|
|
seat->new_drag_icon.notify = handle_new_drag_icon;
|
|
|
|
|
|
|
|
|
|
wl_list_init(&seat->devices);
|
|
|
|
|
|
|
|
|
|
wl_list_insert(&server.input->seats, &seat->link);
|
|
|
|
|
|
|
|
|
|
return seat;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void seat_update_capabilities(struct sway_seat *seat) {
|
|
|
|
|
uint32_t caps = 0;
|
|
|
|
|
struct sway_seat_device *seat_device;
|
|
|
|
|
wl_list_for_each(seat_device, &seat->devices, link) {
|
|
|
|
|
switch (seat_device->input_device->wlr_device->type) {
|
|
|
|
|
case WLR_INPUT_DEVICE_KEYBOARD:
|
|
|
|
|
caps |= WL_SEAT_CAPABILITY_KEYBOARD;
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_POINTER:
|
|
|
|
|
caps |= WL_SEAT_CAPABILITY_POINTER;
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_TOUCH:
|
|
|
|
|
caps |= WL_SEAT_CAPABILITY_TOUCH;
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_TABLET_TOOL:
|
|
|
|
|
caps |= WL_SEAT_CAPABILITY_POINTER;
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_TABLET_PAD:
|
|
|
|
|
case WLR_INPUT_DEVICE_SWITCH:
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Hide cursor if seat doesn't have pointer capability.
|
|
|
|
|
// We must call cursor_set_image while the wlr_seat has the capabilities
|
|
|
|
|
// otherwise it's a no op.
|
|
|
|
|
if ((caps & WL_SEAT_CAPABILITY_POINTER) == 0) {
|
|
|
|
|
cursor_set_image(seat->cursor, NULL, NULL);
|
|
|
|
|
wlr_seat_set_capabilities(seat->wlr_seat, caps);
|
|
|
|
|
} else {
|
|
|
|
|
wlr_seat_set_capabilities(seat->wlr_seat, caps);
|
|
|
|
|
cursor_set_image(seat->cursor, "left_ptr", NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void seat_apply_input_config(struct sway_seat *seat,
|
|
|
|
|
struct sway_seat_device *sway_device) {
|
|
|
|
|
const char *mapped_to_output = NULL;
|
|
|
|
|
|
|
|
|
|
struct input_config *ic = input_device_get_config(
|
|
|
|
|
sway_device->input_device);
|
|
|
|
|
if (ic != NULL) {
|
|
|
|
|
wlr_log(WLR_DEBUG, "Applying input config to %s",
|
|
|
|
|
sway_device->input_device->identifier);
|
|
|
|
|
|
|
|
|
|
mapped_to_output = ic->mapped_to_output;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (mapped_to_output == NULL) {
|
|
|
|
|
mapped_to_output = sway_device->input_device->wlr_device->output_name;
|
|
|
|
|
}
|
|
|
|
|
if (mapped_to_output != NULL) {
|
|
|
|
|
wlr_log(WLR_DEBUG, "Mapping input device %s to output %s",
|
|
|
|
|
sway_device->input_device->identifier, mapped_to_output);
|
|
|
|
|
struct sway_output *output = output_by_name_or_id(mapped_to_output);
|
|
|
|
|
if (output) {
|
|
|
|
|
wlr_cursor_map_input_to_output(seat->cursor->cursor,
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
sway_device->input_device->wlr_device, output->wlr_output);
|
|
|
|
|
wlr_log(WLR_DEBUG, "Mapped to output %s", output->wlr_output->name);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void seat_configure_pointer(struct sway_seat *seat,
|
|
|
|
|
struct sway_seat_device *sway_device) {
|
|
|
|
|
seat_configure_xcursor(seat);
|
|
|
|
|
wlr_cursor_attach_input_device(seat->cursor->cursor,
|
|
|
|
|
sway_device->input_device->wlr_device);
|
|
|
|
|
seat_apply_input_config(seat, sway_device);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void seat_configure_keyboard(struct sway_seat *seat,
|
|
|
|
|
struct sway_seat_device *seat_device) {
|
|
|
|
|
if (!seat_device->keyboard) {
|
|
|
|
|
sway_keyboard_create(seat, seat_device);
|
|
|
|
|
}
|
|
|
|
|
struct wlr_keyboard *wlr_keyboard =
|
|
|
|
|
seat_device->input_device->wlr_device->keyboard;
|
|
|
|
|
sway_keyboard_configure(seat_device->keyboard);
|
|
|
|
|
wlr_seat_set_keyboard(seat->wlr_seat,
|
|
|
|
|
seat_device->input_device->wlr_device);
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *focus = seat_get_focus(seat);
|
|
|
|
|
if (focus && node_is_view(focus)) {
|
|
|
|
|
// force notify reenter to pick up the new configuration
|
|
|
|
|
wlr_seat_keyboard_clear_focus(seat->wlr_seat);
|
|
|
|
|
wlr_seat_keyboard_notify_enter(seat->wlr_seat,
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
focus->sway_container->view->surface, wlr_keyboard->keycodes,
|
|
|
|
|
wlr_keyboard->num_keycodes, &wlr_keyboard->modifiers);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void seat_configure_touch(struct sway_seat *seat,
|
|
|
|
|
struct sway_seat_device *sway_device) {
|
|
|
|
|
wlr_cursor_attach_input_device(seat->cursor->cursor,
|
|
|
|
|
sway_device->input_device->wlr_device);
|
|
|
|
|
seat_apply_input_config(seat, sway_device);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void seat_configure_tablet_tool(struct sway_seat *seat,
|
|
|
|
|
struct sway_seat_device *sway_device) {
|
|
|
|
|
wlr_cursor_attach_input_device(seat->cursor->cursor,
|
|
|
|
|
sway_device->input_device->wlr_device);
|
|
|
|
|
seat_apply_input_config(seat, sway_device);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static struct sway_seat_device *seat_get_device(struct sway_seat *seat,
|
|
|
|
|
struct sway_input_device *input_device) {
|
|
|
|
|
struct sway_seat_device *seat_device = NULL;
|
|
|
|
|
wl_list_for_each(seat_device, &seat->devices, link) {
|
|
|
|
|
if (seat_device->input_device == input_device) {
|
|
|
|
|
return seat_device;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_configure_device(struct sway_seat *seat,
|
|
|
|
|
struct sway_input_device *input_device) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_device *seat_device = seat_get_device(seat, input_device);
|
|
|
|
|
if (!seat_device) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
switch (input_device->wlr_device->type) {
|
|
|
|
|
case WLR_INPUT_DEVICE_POINTER:
|
|
|
|
|
seat_configure_pointer(seat, seat_device);
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_KEYBOARD:
|
|
|
|
|
seat_configure_keyboard(seat, seat_device);
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_TOUCH:
|
|
|
|
|
seat_configure_touch(seat, seat_device);
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_TABLET_TOOL:
|
|
|
|
|
seat_configure_tablet_tool(seat, seat_device);
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_TABLET_PAD:
|
|
|
|
|
wlr_log(WLR_DEBUG, "TODO: configure tablet pad");
|
|
|
|
|
break;
|
|
|
|
|
case WLR_INPUT_DEVICE_SWITCH:
|
|
|
|
|
wlr_log(WLR_DEBUG, "TODO: configure switch device");
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_add_device(struct sway_seat *seat,
|
|
|
|
|
struct sway_input_device *input_device) {
|
|
|
|
|
if (seat_get_device(seat, input_device)) {
|
|
|
|
|
seat_configure_device(seat, input_device);
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_seat_device *seat_device =
|
|
|
|
|
calloc(1, sizeof(struct sway_seat_device));
|
|
|
|
|
if (!seat_device) {
|
|
|
|
|
wlr_log(WLR_DEBUG, "could not allocate seat device");
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
wlr_log(WLR_DEBUG, "adding device %s to seat %s",
|
|
|
|
|
input_device->identifier, seat->wlr_seat->name);
|
|
|
|
|
|
|
|
|
|
seat_device->sway_seat = seat;
|
|
|
|
|
seat_device->input_device = input_device;
|
|
|
|
|
wl_list_insert(&seat->devices, &seat_device->link);
|
|
|
|
|
|
|
|
|
|
seat_configure_device(seat, input_device);
|
|
|
|
|
|
|
|
|
|
seat_update_capabilities(seat);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_remove_device(struct sway_seat *seat,
|
|
|
|
|
struct sway_input_device *input_device) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_device *seat_device = seat_get_device(seat, input_device);
|
|
|
|
|
|
|
|
|
|
if (!seat_device) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
wlr_log(WLR_DEBUG, "removing device %s from seat %s",
|
|
|
|
|
input_device->identifier, seat->wlr_seat->name);
|
|
|
|
|
|
|
|
|
|
seat_device_destroy(seat_device);
|
|
|
|
|
|
|
|
|
|
seat_update_capabilities(seat);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_configure_xcursor(struct sway_seat *seat) {
|
|
|
|
|
// TODO configure theme and size
|
|
|
|
|
const char *cursor_theme = NULL;
|
|
|
|
|
|
|
|
|
|
if (!seat->cursor->xcursor_manager) {
|
|
|
|
|
seat->cursor->xcursor_manager =
|
|
|
|
|
wlr_xcursor_manager_create(cursor_theme, 24);
|
|
|
|
|
if (sway_assert(seat->cursor->xcursor_manager,
|
|
|
|
|
"Cannot create XCursor manager for theme %s",
|
|
|
|
|
cursor_theme)) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
for (int i = 0; i < root->outputs->length; ++i) {
|
|
|
|
|
struct sway_output *sway_output = root->outputs->items[i];
|
|
|
|
|
struct wlr_output *output = sway_output->wlr_output;
|
|
|
|
|
bool result =
|
|
|
|
|
wlr_xcursor_manager_load(seat->cursor->xcursor_manager,
|
|
|
|
|
output->scale);
|
|
|
|
|
|
|
|
|
|
sway_assert(!result,
|
|
|
|
|
"Cannot load xcursor theme for output '%s' with scale %f",
|
|
|
|
|
// TODO: Fractional scaling
|
|
|
|
|
output->name, (double)output->scale);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
cursor_set_image(seat->cursor, "left_ptr", NULL);
|
|
|
|
|
wlr_cursor_warp(seat->cursor->cursor, NULL, seat->cursor->cursor->x,
|
|
|
|
|
seat->cursor->cursor->y);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bool seat_is_input_allowed(struct sway_seat *seat,
|
|
|
|
|
struct wlr_surface *surface) {
|
|
|
|
|
struct wl_client *client = wl_resource_get_client(surface->resource);
|
|
|
|
|
return !seat->exclusive_client || seat->exclusive_client == client;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void send_unfocus(struct sway_container *con, void *data) {
|
|
|
|
|
if (con->view) {
|
|
|
|
|
view_set_activated(con->view, false);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Unfocus the container and any children (eg. when leaving `focus parent`)
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
static void seat_send_unfocus(struct sway_node *node, struct sway_seat *seat) {
|
|
|
|
|
wlr_seat_keyboard_clear_focus(seat->wlr_seat);
|
|
|
|
|
if (node->type == N_WORKSPACE) {
|
|
|
|
|
workspace_for_each_container(node->sway_workspace, send_unfocus, seat);
|
|
|
|
|
} else {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
send_unfocus(node->sway_container, seat);
|
|
|
|
|
container_for_each_child(node->sway_container, send_unfocus, seat);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static int handle_urgent_timeout(void *data) {
|
|
|
|
|
struct sway_view *view = data;
|
|
|
|
|
view_set_urgent(view, false);
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void container_raise_floating(struct sway_container *con) {
|
|
|
|
|
// Bring container to front by putting it at the end of the floating list.
|
|
|
|
|
struct sway_container *floater = con;
|
|
|
|
|
while (floater->parent) {
|
|
|
|
|
floater = floater->parent;
|
|
|
|
|
}
|
|
|
|
|
if (container_is_floating(floater)) {
|
|
|
|
|
list_move_to_end(floater->workspace->floating, floater);
|
|
|
|
|
node_set_dirty(&floater->workspace->node);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void set_workspace(struct sway_seat *seat,
|
|
|
|
|
struct sway_workspace *new_ws) {
|
|
|
|
|
if (seat->workspace == new_ws) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
ipc_event_workspace(seat->workspace, new_ws, "focus");
|
|
|
|
|
seat->workspace = new_ws;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_set_raw_focus(struct sway_seat *seat, struct sway_node *node) {
|
|
|
|
|
struct sway_seat_node *seat_node = seat_node_from_node(seat, node);
|
|
|
|
|
wl_list_remove(&seat_node->link);
|
|
|
|
|
wl_list_insert(&seat->focus_stack, &seat_node->link);
|
|
|
|
|
node_set_dirty(node);
|
|
|
|
|
node_set_dirty(node_get_parent(node));
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_set_focus(struct sway_seat *seat, struct sway_node *node) {
|
|
|
|
|
if (seat->focused_layer) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *last_focus = seat_get_focus(seat);
|
|
|
|
|
if (last_focus == node) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_workspace *last_workspace = seat_get_focused_workspace(seat);
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (node == NULL) {
|
|
|
|
|
// Close any popups on the old focus
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (node_is_view(last_focus)) {
|
|
|
|
|
view_close_popups(last_focus->sway_container->view);
|
|
|
|
|
}
|
|
|
|
|
seat_send_unfocus(last_focus, seat);
|
|
|
|
|
seat->has_focus = false;
|
|
|
|
|
update_debug_tree();
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_workspace *new_workspace = node->type == N_WORKSPACE ?
|
|
|
|
|
node->sway_workspace : node->sway_container->workspace;
|
|
|
|
|
struct sway_container *container = node->type == N_CONTAINER ?
|
|
|
|
|
node->sway_container : NULL;
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
// Deny setting focus to a view which is hidden by a fullscreen container
|
|
|
|
|
if (new_workspace && new_workspace->fullscreen && container &&
|
|
|
|
|
!container_is_fullscreen_or_child(container)) {
|
|
|
|
|
// Unless it's a transient container
|
|
|
|
|
if (!container_is_transient_for(container, new_workspace->fullscreen)) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_output *new_output = new_workspace->output;
|
|
|
|
|
|
|
|
|
|
if (last_workspace != new_workspace && new_output) {
|
|
|
|
|
node_set_dirty(&new_output->node);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// find new output's old workspace, which might have to be removed if empty
|
|
|
|
|
struct sway_workspace *new_output_last_ws =
|
|
|
|
|
new_output ? output_get_active_workspace(new_output) : NULL;
|
|
|
|
|
|
|
|
|
|
// Unfocus the previous focus
|
|
|
|
|
if (last_focus) {
|
|
|
|
|
seat_send_unfocus(last_focus, seat);
|
|
|
|
|
node_set_dirty(last_focus);
|
|
|
|
|
struct sway_node *parent = node_get_parent(last_focus);
|
|
|
|
|
if (parent) {
|
|
|
|
|
node_set_dirty(parent);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
// Put the container parents on the focus stack, then the workspace, then
|
|
|
|
|
// the focused container.
|
|
|
|
|
if (container) {
|
|
|
|
|
struct sway_container *parent = container->parent;
|
|
|
|
|
while (parent) {
|
|
|
|
|
seat_set_raw_focus(seat, &parent->node);
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
parent = parent->parent;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
}
|
|
|
|
|
if (new_workspace) {
|
|
|
|
|
seat_set_raw_focus(seat, &new_workspace->node);
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
}
|
|
|
|
|
if (container) {
|
|
|
|
|
seat_set_raw_focus(seat, &container->node);
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
seat_send_focus(&container->node, seat);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// emit ipc events
|
|
|
|
|
set_workspace(seat, new_workspace);
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (container && container->view) {
|
|
|
|
|
ipc_event_window(container, "focus");
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Move sticky containers to new workspace
|
|
|
|
|
if (new_output_last_ws && new_workspace != new_output_last_ws) {
|
|
|
|
|
for (int i = 0; i < new_output_last_ws->floating->length; ++i) {
|
|
|
|
|
struct sway_container *floater =
|
|
|
|
|
new_output_last_ws->floating->items[i];
|
|
|
|
|
if (floater->is_sticky) {
|
|
|
|
|
container_detach(floater);
|
|
|
|
|
workspace_add_floating(new_workspace, floater);
|
|
|
|
|
--i;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Close any popups on the old focus
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (last_focus && node_is_view(last_focus)) {
|
|
|
|
|
view_close_popups(last_focus->sway_container->view);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// If urgent, either unset the urgency or start a timer to unset it
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (container && container->view && view_is_urgent(container->view) &&
|
|
|
|
|
!container->view->urgent_timer) {
|
|
|
|
|
struct sway_view *view = container->view;
|
|
|
|
|
if (last_workspace && last_workspace != new_workspace &&
|
|
|
|
|
config->urgent_timeout > 0) {
|
|
|
|
|
view->urgent_timer = wl_event_loop_add_timer(server.wl_event_loop,
|
|
|
|
|
handle_urgent_timeout, view);
|
|
|
|
|
if (view->urgent_timer) {
|
|
|
|
|
wl_event_source_timer_update(view->urgent_timer,
|
|
|
|
|
config->urgent_timeout);
|
|
|
|
|
} else {
|
|
|
|
|
wlr_log(WLR_ERROR, "Unable to create urgency timer (%s)",
|
|
|
|
|
strerror(errno));
|
|
|
|
|
handle_urgent_timeout(view);
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
view_set_urgent(view, false);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (new_output_last_ws) {
|
|
|
|
|
workspace_consider_destroy(new_output_last_ws);
|
|
|
|
|
}
|
|
|
|
|
if (last_workspace && last_workspace != new_output_last_ws) {
|
|
|
|
|
workspace_consider_destroy(last_workspace);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
seat->has_focus = true;
|
|
|
|
|
|
|
|
|
|
if (config->smart_gaps) {
|
|
|
|
|
// When smart gaps is on, gaps may change when the focus changes so
|
|
|
|
|
// the workspace needs to be arranged
|
|
|
|
|
arrange_workspace(new_workspace);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
update_debug_tree();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_set_focus_container(struct sway_seat *seat,
|
|
|
|
|
struct sway_container *con) {
|
|
|
|
|
seat_set_focus(seat, con ? &con->node : NULL);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_set_focus_workspace(struct sway_seat *seat,
|
|
|
|
|
struct sway_workspace *ws) {
|
|
|
|
|
seat_set_focus(seat, ws ? &ws->node : NULL);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_set_focus_surface(struct sway_seat *seat,
|
|
|
|
|
struct wlr_surface *surface, bool unfocus) {
|
|
|
|
|
if (seat->has_focus && unfocus) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *focus = seat_get_focus(seat);
|
|
|
|
|
seat_send_unfocus(focus, seat);
|
|
|
|
|
seat->has_focus = false;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct wlr_keyboard *keyboard = wlr_seat_get_keyboard(seat->wlr_seat);
|
|
|
|
|
if (keyboard) {
|
|
|
|
|
wlr_seat_keyboard_notify_enter(seat->wlr_seat, surface,
|
|
|
|
|
keyboard->keycodes, keyboard->num_keycodes, &keyboard->modifiers);
|
|
|
|
|
} else {
|
|
|
|
|
wlr_seat_keyboard_notify_enter(seat->wlr_seat, surface, NULL, 0, NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_set_focus_layer(struct sway_seat *seat,
|
|
|
|
|
struct wlr_layer_surface_v1 *layer) {
|
|
|
|
|
if (!layer && seat->focused_layer) {
|
|
|
|
|
seat->focused_layer = NULL;
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *previous = seat_get_focus_inactive(seat, &root->node);
|
|
|
|
|
if (previous) {
|
|
|
|
|
// Hack to get seat to re-focus the return value of get_focus
|
|
|
|
|
seat_set_focus(seat, NULL);
|
|
|
|
|
seat_set_focus(seat, previous);
|
|
|
|
|
}
|
|
|
|
|
return;
|
|
|
|
|
} else if (!layer || seat->focused_layer == layer) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
seat_set_focus_surface(seat, layer->surface, true);
|
|
|
|
|
if (layer->layer >= ZWLR_LAYER_SHELL_V1_LAYER_TOP) {
|
|
|
|
|
seat->focused_layer = layer;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_set_exclusive_client(struct sway_seat *seat,
|
|
|
|
|
struct wl_client *client) {
|
|
|
|
|
if (!client) {
|
|
|
|
|
seat->exclusive_client = client;
|
|
|
|
|
// Triggers a refocus of the topmost surface layer if necessary
|
|
|
|
|
// TODO: Make layer surface focus per-output based on cursor position
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
for (int i = 0; i < root->outputs->length; ++i) {
|
|
|
|
|
struct sway_output *output = root->outputs->items[i];
|
|
|
|
|
arrange_layers(output);
|
|
|
|
|
}
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (seat->focused_layer) {
|
|
|
|
|
if (wl_resource_get_client(seat->focused_layer->resource) != client) {
|
|
|
|
|
seat_set_focus_layer(seat, NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (seat->has_focus) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *focus = seat_get_focus(seat);
|
|
|
|
|
if (node_is_view(focus) && wl_resource_get_client(
|
|
|
|
|
focus->sway_container->view->surface->resource) != client) {
|
|
|
|
|
seat_set_focus(seat, NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (seat->wlr_seat->pointer_state.focused_client) {
|
|
|
|
|
if (seat->wlr_seat->pointer_state.focused_client->client != client) {
|
|
|
|
|
wlr_seat_pointer_clear_focus(seat->wlr_seat);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
struct timespec now;
|
|
|
|
|
clock_gettime(CLOCK_MONOTONIC, &now);
|
|
|
|
|
struct wlr_touch_point *point;
|
|
|
|
|
wl_list_for_each(point, &seat->wlr_seat->touch_state.touch_points, link) {
|
|
|
|
|
if (point->client->client != client) {
|
|
|
|
|
wlr_seat_touch_point_clear_focus(seat->wlr_seat,
|
|
|
|
|
now.tv_nsec / 1000, point->touch_id);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
seat->exclusive_client = client;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *seat_get_focus_inactive(struct sway_seat *seat,
|
|
|
|
|
struct sway_node *node) {
|
|
|
|
|
if (node_is_view(node)) {
|
|
|
|
|
return node;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *current;
|
|
|
|
|
wl_list_for_each(current, &seat->focus_stack, link) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (node_has_ancestor(current->node, node)) {
|
|
|
|
|
return current->node;
|
|
|
|
|
}
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
if (node->type == N_WORKSPACE) {
|
|
|
|
|
return node;
|
|
|
|
|
}
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_container *seat_get_focus_inactive_tiling(struct sway_seat *seat,
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_workspace *workspace) {
|
|
|
|
|
if (!workspace->tiling->length) {
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *current;
|
|
|
|
|
wl_list_for_each(current, &seat->focus_stack, link) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *node = current->node;
|
|
|
|
|
if (node->type == N_CONTAINER &&
|
|
|
|
|
!container_is_floating_or_child(node->sway_container) &&
|
|
|
|
|
node->sway_container->workspace == workspace) {
|
|
|
|
|
return node->sway_container;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_container *seat_get_focus_inactive_floating(struct sway_seat *seat,
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_workspace *workspace) {
|
|
|
|
|
if (!workspace->floating->length) {
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *current;
|
|
|
|
|
wl_list_for_each(current, &seat->focus_stack, link) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *node = current->node;
|
|
|
|
|
if (node->type == N_CONTAINER &&
|
|
|
|
|
container_is_floating_or_child(node->sway_container) &&
|
|
|
|
|
node->sway_container->workspace == workspace) {
|
|
|
|
|
return node->sway_container;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_node *seat_get_active_tiling_child(struct sway_seat *seat,
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *parent) {
|
|
|
|
|
if (node_is_view(parent)) {
|
|
|
|
|
return parent;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *current;
|
|
|
|
|
wl_list_for_each(current, &seat->focus_stack, link) {
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *node = current->node;
|
|
|
|
|
if (node_get_parent(node) != parent) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
if (parent->type == N_WORKSPACE) {
|
|
|
|
|
// Only consider tiling children
|
|
|
|
|
struct sway_workspace *ws = parent->sway_workspace;
|
|
|
|
|
if (list_find(ws->tiling, node->sway_container) == -1) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return node;
|
|
|
|
|
}
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_node *seat_get_focus(struct sway_seat *seat) {
|
|
|
|
|
if (!seat->has_focus) {
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
if (wl_list_empty(&seat->focus_stack)) {
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
struct sway_seat_node *current =
|
|
|
|
|
wl_container_of(seat->focus_stack.next, current, link);
|
Implement type safe arguments and demote sway_container
This commit changes the meaning of sway_container so that it only refers
to layout containers and view containers. Workspaces, outputs and the
root are no longer known as containers. Instead, root, outputs,
workspaces and containers are all a type of node, and containers come in
two types: layout containers and view containers.
In addition to the above, this implements type safe variables. This
means we use specific types such as sway_output and sway_workspace
instead of generic containers or nodes. However, it's worth noting that
in a few places places (eg. seat focus and transactions) referring to
them in a generic way is unavoidable which is why we still use nodes in
some places.
If you want a TL;DR, look at node.h, as well as the struct definitions
for root, output, workspace and container. Note that sway_output now
contains a workspaces list, and workspaces now contain a tiling and
floating list, and containers now contain a pointer back to the
workspace.
There are now functions for seat_get_focused_workspace and
seat_get_focused_container. The latter will return NULL if a workspace
itself is focused. Most other seat functions like seat_get_focus and
seat_set_focus now accept and return nodes.
In the config->handler_context struct, current_container has been
replaced with three pointers: node, container and workspace. node is the
same as what current_container was, while workspace is the workspace
that the node resides on and container is the actual container, which
may be NULL if a workspace itself is focused.
The global root_container variable has been replaced with one simply
called root, which is a pointer to the sway_root instance.
The way outputs are created, enabled, disabled and destroyed has
changed. Previously we'd wrap the sway_output in a container when it is
enabled, but as we don't have containers any more it needs a different
approach. The output_create and output_destroy functions previously
created/destroyed the container, but now they create/destroy the
sway_output. There is a new function output_disable to disable an output
without destroying it.
Containers have a new view property. If this is populated then the
container is a view container, otherwise it's a layout container. Like
before, this property is immutable for the life of the container.
Containers have both a `sway_container *parent` and
`sway_workspace *workspace`. As we use specific types now, parent cannot
point to a workspace so it'll be NULL for containers which are direct
children of the workspace. The workspace property is set for all
containers, except those which are hidden in the scratchpad as they have
no workspace.
In some cases we need to refer to workspaces in a container-like way.
For example, workspaces have layout and children, but when using
specific types this makes it difficult. Likewise, it's difficult for a
container to get its parent's layout when the parent could be another
container or a workspace. To make it easier, some helper functions have
been created: container_parent_layout and container_get_siblings.
container_remove_child has been renamed to container_detach and
container_replace_child has been renamed to container_replace.
`container_handle_fullscreen_reparent(con, old_parent)` has had the
old_parent removed. We now unfullscreen the workspace when detaching the
container, so this function is simplified and only needs one argument
now.
container_notify_subtree_changed has been renamed to
container_update_representation. This is more descriptive of its
purpose. I also wanted to be able to call it with whatever container was
changed rather than the container's parent, which makes bubbling up to
the workspace easier.
There are now state structs per node thing. ie. sway_output_state,
sway_workspace_state and sway_container_state.
The focus, move and layout commands have been completely refactored to
work with the specific types. I considered making these a separate PR,
but I'd be backporting my changes only to replace them again, and it's
easier just to test everything at once.
6 years ago
|
|
|
return current->node;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_workspace *seat_get_focused_workspace(struct sway_seat *seat) {
|
|
|
|
|
struct sway_node *focus = seat_get_focus(seat);
|
|
|
|
|
if (!focus) {
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
if (focus->type == N_CONTAINER) {
|
|
|
|
|
return focus->sway_container->workspace;
|
|
|
|
|
}
|
|
|
|
|
if (focus->type == N_WORKSPACE) {
|
|
|
|
|
return focus->sway_workspace;
|
|
|
|
|
}
|
|
|
|
|
return NULL; // unreachable
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct sway_container *seat_get_focused_container(struct sway_seat *seat) {
|
|
|
|
|
struct sway_node *focus = seat_get_focus(seat);
|
|
|
|
|
if (focus && focus->type == N_CONTAINER) {
|
|
|
|
|
return focus->sway_container;
|
|
|
|
|
}
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_apply_config(struct sway_seat *seat,
|
|
|
|
|
struct seat_config *seat_config) {
|
|
|
|
|
struct sway_seat_device *seat_device = NULL;
|
|
|
|
|
|
|
|
|
|
if (!seat_config) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
wl_list_for_each(seat_device, &seat->devices, link) {
|
|
|
|
|
seat_configure_device(seat, seat_device->input_device);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
cursor_handle_activity(seat->cursor);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct seat_config *seat_get_config(struct sway_seat *seat) {
|
|
|
|
|
struct seat_config *seat_config = NULL;
|
|
|
|
|
for (int i = 0; i < config->seat_configs->length; ++i ) {
|
|
|
|
|
seat_config = config->seat_configs->items[i];
|
|
|
|
|
if (strcmp(seat->wlr_seat->name, seat_config->name) == 0) {
|
|
|
|
|
return seat_config;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
struct seat_config *seat_get_config_by_name(const char *name) {
|
|
|
|
|
struct seat_config *seat_config = NULL;
|
|
|
|
|
for (int i = 0; i < config->seat_configs->length; ++i ) {
|
|
|
|
|
seat_config = config->seat_configs->items[i];
|
|
|
|
|
if (strcmp(name, seat_config->name) == 0) {
|
|
|
|
|
return seat_config;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_begin_down(struct sway_seat *seat, struct sway_container *con,
|
|
|
|
|
uint32_t button, double sx, double sy) {
|
|
|
|
|
seat->operation = OP_DOWN;
|
|
|
|
|
seat->op_container = con;
|
|
|
|
|
seat->op_button = button;
|
|
|
|
|
seat->op_ref_lx = seat->cursor->cursor->x;
|
|
|
|
|
seat->op_ref_ly = seat->cursor->cursor->y;
|
|
|
|
|
seat->op_ref_con_lx = sx;
|
|
|
|
|
seat->op_ref_con_ly = sy;
|
|
|
|
|
seat->op_moved = false;
|
|
|
|
|
|
|
|
|
|
container_raise_floating(con);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_begin_move_floating(struct sway_seat *seat,
|
|
|
|
|
struct sway_container *con, uint32_t button) {
|
|
|
|
|
if (!seat->cursor) {
|
|
|
|
|
wlr_log(WLR_DEBUG, "Ignoring move request due to no cursor device");
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
seat->operation = OP_MOVE_FLOATING;
|
|
|
|
|
seat->op_container = con;
|
|
|
|
|
seat->op_button = button;
|
|
|
|
|
|
|
|
|
|
container_raise_floating(con);
|
|
|
|
|
|
|
|
|
|
cursor_set_image(seat->cursor, "grab", NULL);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_begin_move_tiling_threshold(struct sway_seat *seat,
|
|
|
|
|
struct sway_container *con, uint32_t button) {
|
|
|
|
|
seat->operation = OP_MOVE_TILING_THRESHOLD;
|
|
|
|
|
seat->op_container = con;
|
|
|
|
|
seat->op_button = button;
|
|
|
|
|
seat->op_target_node = NULL;
|
|
|
|
|
seat->op_target_edge = 0;
|
|
|
|
|
seat->op_ref_lx = seat->cursor->cursor->x;
|
|
|
|
|
seat->op_ref_ly = seat->cursor->cursor->y;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_begin_move_tiling(struct sway_seat *seat,
|
|
|
|
|
struct sway_container *con, uint32_t button) {
|
|
|
|
|
seat->operation = OP_MOVE_TILING;
|
|
|
|
|
seat->op_container = con;
|
|
|
|
|
seat->op_button = button;
|
|
|
|
|
seat->op_target_node = NULL;
|
|
|
|
|
seat->op_target_edge = 0;
|
|
|
|
|
cursor_set_image(seat->cursor, "grab", NULL);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_begin_resize_floating(struct sway_seat *seat,
|
|
|
|
|
struct sway_container *con, uint32_t button, enum wlr_edges edge) {
|
|
|
|
|
if (!seat->cursor) {
|
|
|
|
|
wlr_log(WLR_DEBUG, "Ignoring resize request due to no cursor device");
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
struct wlr_keyboard *keyboard = wlr_seat_get_keyboard(seat->wlr_seat);
|
|
|
|
|
seat->operation = OP_RESIZE_FLOATING;
|
|
|
|
|
seat->op_container = con;
|
|
|
|
|
seat->op_resize_preserve_ratio = keyboard &&
|
|
|
|
|
(wlr_keyboard_get_modifiers(keyboard) & WLR_MODIFIER_SHIFT);
|
|
|
|
|
seat->op_resize_edge = edge == WLR_EDGE_NONE ?
|
|
|
|
|
WLR_EDGE_BOTTOM | WLR_EDGE_RIGHT : edge;
|
|
|
|
|
seat->op_button = button;
|
|
|
|
|
seat->op_ref_lx = seat->cursor->cursor->x;
|
|
|
|
|
seat->op_ref_ly = seat->cursor->cursor->y;
|
|
|
|
|
seat->op_ref_con_lx = con->x;
|
|
|
|
|
seat->op_ref_con_ly = con->y;
|
|
|
|
|
seat->op_ref_width = con->width;
|
|
|
|
|
seat->op_ref_height = con->height;
|
|
|
|
|
|
|
|
|
|
container_raise_floating(con);
|
|
|
|
|
|
|
|
|
|
const char *image = edge == WLR_EDGE_NONE ?
|
|
|
|
|
"se-resize" : wlr_xcursor_get_resize_name(edge);
|
|
|
|
|
cursor_set_image(seat->cursor, image, NULL);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_begin_resize_tiling(struct sway_seat *seat,
|
|
|
|
|
struct sway_container *con, uint32_t button, enum wlr_edges edge) {
|
|
|
|
|
seat->operation = OP_RESIZE_TILING;
|
|
|
|
|
seat->op_container = con;
|
|
|
|
|
seat->op_resize_edge = edge;
|
|
|
|
|
seat->op_button = button;
|
|
|
|
|
seat->op_ref_lx = seat->cursor->cursor->x;
|
|
|
|
|
seat->op_ref_ly = seat->cursor->cursor->y;
|
|
|
|
|
seat->op_ref_con_lx = con->x;
|
|
|
|
|
seat->op_ref_con_ly = con->y;
|
|
|
|
|
seat->op_ref_width = con->width;
|
|
|
|
|
seat->op_ref_height = con->height;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static bool is_parallel(enum sway_container_layout layout,
|
|
|
|
|
enum wlr_edges edge) {
|
|
|
|
|
bool layout_is_horiz = layout == L_HORIZ || layout == L_TABBED;
|
|
|
|
|
bool edge_is_horiz = edge == WLR_EDGE_LEFT || edge == WLR_EDGE_RIGHT;
|
|
|
|
|
return layout_is_horiz == edge_is_horiz;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void seat_end_move_tiling(struct sway_seat *seat) {
|
|
|
|
|
struct sway_container *con = seat->op_container;
|
|
|
|
|
struct sway_container *old_parent = con->parent;
|
|
|
|
|
struct sway_workspace *old_ws = con->workspace;
|
|
|
|
|
struct sway_node *target_node = seat->op_target_node;
|
|
|
|
|
struct sway_workspace *new_ws = target_node->type == N_WORKSPACE ?
|
|
|
|
|
target_node->sway_workspace : target_node->sway_container->workspace;
|
|
|
|
|
enum wlr_edges edge = seat->op_target_edge;
|
|
|
|
|
int after = edge != WLR_EDGE_TOP && edge != WLR_EDGE_LEFT;
|
|
|
|
|
|
|
|
|
|
container_detach(con);
|
|
|
|
|
|
|
|
|
|
// Moving container into empty workspace
|
|
|
|
|
if (target_node->type == N_WORKSPACE && edge == WLR_EDGE_NONE) {
|
|
|
|
|
workspace_add_tiling(new_ws, con);
|
|
|
|
|
} else if (target_node->type == N_CONTAINER) {
|
|
|
|
|
// Moving container before/after another
|
|
|
|
|
struct sway_container *target = target_node->sway_container;
|
|
|
|
|
enum sway_container_layout layout = container_parent_layout(target);
|
|
|
|
|
if (edge && !is_parallel(layout, edge)) {
|
|
|
|
|
enum sway_container_layout new_layout = edge == WLR_EDGE_TOP ||
|
|
|
|
|
edge == WLR_EDGE_BOTTOM ? L_VERT : L_HORIZ;
|
|
|
|
|
container_split(target, new_layout);
|
|
|
|
|
}
|
|
|
|
|
container_add_sibling(target, con, after);
|
|
|
|
|
} else {
|
|
|
|
|
// Target is a workspace which requires splitting
|
|
|
|
|
enum sway_container_layout new_layout = edge == WLR_EDGE_TOP ||
|
|
|
|
|
edge == WLR_EDGE_BOTTOM ? L_VERT : L_HORIZ;
|
|
|
|
|
workspace_split(new_ws, new_layout);
|
|
|
|
|
workspace_insert_tiling(new_ws, con, after);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (old_parent) {
|
|
|
|
|
container_reap_empty(old_parent);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// This is a bit dirty, but we'll set the dimensions to that of a sibling.
|
|
|
|
|
// I don't think there's any other way to make it consistent without
|
|
|
|
|
// changing how we auto-size containers.
|
|
|
|
|
list_t *siblings = container_get_siblings(con);
|
|
|
|
|
if (siblings->length > 1) {
|
|
|
|
|
int index = list_find(siblings, con);
|
|
|
|
|
struct sway_container *sibling = index == 0 ?
|
|
|
|
|
siblings->items[1] : siblings->items[index - 1];
|
|
|
|
|
con->width = sibling->width;
|
|
|
|
|
con->height = sibling->height;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
arrange_workspace(old_ws);
|
|
|
|
|
if (new_ws != old_ws) {
|
|
|
|
|
arrange_workspace(new_ws);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_end_mouse_operation(struct sway_seat *seat) {
|
|
|
|
|
enum sway_seat_operation operation = seat->operation;
|
|
|
|
|
if (seat->operation == OP_MOVE_FLOATING) {
|
|
|
|
|
// We "move" the container to its own location so it discovers its
|
|
|
|
|
// output again.
|
|
|
|
|
struct sway_container *con = seat->op_container;
|
|
|
|
|
container_floating_move_to(con, con->x, con->y);
|
|
|
|
|
} else if (seat->operation == OP_MOVE_TILING && seat->op_target_node) {
|
|
|
|
|
seat_end_move_tiling(seat);
|
|
|
|
|
}
|
|
|
|
|
seat->operation = OP_NONE;
|
|
|
|
|
seat->op_container = NULL;
|
|
|
|
|
if (operation == OP_DOWN) {
|
|
|
|
|
// Set the cursor's previous coords to the x/y at the start of the
|
|
|
|
|
// operation, so the container change will be detected if using
|
|
|
|
|
// focus_follows_mouse and the cursor moved off the original container
|
|
|
|
|
// during the operation.
|
|
|
|
|
seat->cursor->previous.x = seat->op_ref_lx;
|
|
|
|
|
seat->cursor->previous.y = seat->op_ref_ly;
|
|
|
|
|
if (seat->op_moved) {
|
|
|
|
|
cursor_send_pointer_motion(seat->cursor, 0);
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
cursor_set_image(seat->cursor, "left_ptr", NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_pointer_notify_button(struct sway_seat *seat, uint32_t time_msec,
|
|
|
|
|
uint32_t button, enum wlr_button_state state) {
|
|
|
|
|
seat->last_button = button;
|
|
|
|
|
seat->last_button_serial = wlr_seat_pointer_notify_button(seat->wlr_seat,
|
|
|
|
|
time_msec, button, state);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void seat_consider_warp_to_focus(struct sway_seat *seat) {
|
|
|
|
|
struct sway_node *focus = seat_get_focus(seat);
|
|
|
|
|
if (config->mouse_warping == WARP_NO || !focus) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
if (config->mouse_warping == WARP_OUTPUT) {
|
|
|
|
|
struct sway_output *output = node_get_output(focus);
|
|
|
|
|
struct wlr_box box;
|
|
|
|
|
output_get_box(output, &box);
|
|
|
|
|
if (wlr_box_contains_point(&box,
|
|
|
|
|
seat->cursor->cursor->x, seat->cursor->cursor->y)) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (focus->type == N_CONTAINER) {
|
|
|
|
|
cursor_warp_to_container(seat->cursor, focus->sway_container);
|
|
|
|
|
} else {
|
|
|
|
|
cursor_warp_to_workspace(seat->cursor, focus->sway_workspace);
|
|
|
|
|
}
|
|
|
|
|
if (seat->cursor->hidden){
|
|
|
|
|
cursor_unhide(seat->cursor);
|
|
|
|
|
wl_event_source_timer_update(seat->cursor->hide_source, cursor_get_timeout(seat->cursor));
|
|
|
|
|
}
|
|
|
|
|
}
|
