itycodes
/
convince
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '20deb8ec16'
${ noResults }
convince/sway/input/seatop_default.c

748 lines
24 KiB
Raw Normal View History Unescape

Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
#define _POSIX_C_SOURCE 200809L
#include <float.h>
#include <libevdev/libevdev.h>
#include <wlr/types/wlr_cursor.h>
input: refactor tablet motion into seatop handler This commit moves tablet motion logic into a seatop handler. As a side-effect of seatop implementations being able to receive tablet motion events, fixes #5232.
5 years ago
#include <wlr/types/wlr_tablet_v2.h>
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
#include <wlr/types/wlr_xcursor_manager.h>
#include "sway/input/cursor.h"
#include "sway/input/seat.h"
input: refactor tablet motion into seatop handler This commit moves tablet motion logic into a seatop handler. As a side-effect of seatop implementations being able to receive tablet motion events, fixes #5232.
5 years ago
#include "sway/input/tablet.h"
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
#include "sway/tree/view.h"
#include "log.h"
seatop_default: handle focus for xwayland_unmanaged views
5 years ago
#if HAVE_XWAYLAND
#include "sway/xwayland.h"
#endif
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
struct seatop_default_event {
struct sway_node *previous_node;
uint32_t pressed_buttons[SWAY_CURSOR_PRESSED_BUTTONS_CAP];
size_t pressed_button_count;
};
/*-----------------------------------------\
* Functions shared by multiple callbacks /
*---------------------------------------*/
/**
* Determine if the edge of the given container is on the edge of the
* workspace/output.
*/
static bool edge_is_external(struct sway_container *cont, enum wlr_edges edge) {
enum sway_container_layout layout = L_NONE;
switch (edge) {
case WLR_EDGE_TOP:
case WLR_EDGE_BOTTOM:
layout = L_VERT;
break;
case WLR_EDGE_LEFT:
case WLR_EDGE_RIGHT:
layout = L_HORIZ;
break;
case WLR_EDGE_NONE:
sway_assert(false, "Never reached");
return false;
}
// Iterate the parents until we find one with the layout we want,
// then check if the child has siblings between it and the edge.
while (cont) {
if (container_parent_layout(cont) == layout) {
list_t *siblings = container_get_siblings(cont);
int index = list_find(siblings, cont);
if (index > 0 && (edge == WLR_EDGE_LEFT || edge == WLR_EDGE_TOP)) {
return false;
}
if (index < siblings->length - 1 &&
(edge == WLR_EDGE_RIGHT || edge == WLR_EDGE_BOTTOM)) {
return false;
}
}
cont = cont->parent;
}
return true;
}
static enum wlr_edges find_edge(struct sway_container *cont,
Do not search for edges on subsurfaces, fix #4381 Subsurfaces (in most cases popups) aren't decorated by sway and will never have any borders, but may be drawn beyond container boundaries producing false positive when searching for edge. So we want to skip edge search when handling mouse event on subsurface.
5 years ago
struct wlr_surface *surface, struct sway_cursor *cursor) {
Fix tiled containers resize with mouse Fixes regression introduced in 2c1a11016ccb7646a74de51eff003c18e5aa7902
5 years ago
if (!cont->view || (surface && cont->view->surface != surface)) {
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
return WLR_EDGE_NONE;
}
if (cont->border == B_NONE || !cont->border_thickness ||
cont->border == B_CSD) {
return WLR_EDGE_NONE;
}
enum wlr_edges edge = 0;
if (cursor->cursor->x < cont->x + cont->border_thickness) {
edge |= WLR_EDGE_LEFT;
}
if (cursor->cursor->y < cont->y + cont->border_thickness) {
edge |= WLR_EDGE_TOP;
}
if (cursor->cursor->x >= cont->x + cont->width - cont->border_thickness) {
edge |= WLR_EDGE_RIGHT;
}
if (cursor->cursor->y >= cont->y + cont->height - cont->border_thickness) {
edge |= WLR_EDGE_BOTTOM;
}
return edge;
}
/**
* If the cursor is over a _resizable_ edge, return the edge.
* Edges that can't be resized are edges of the workspace.
*/
input/cursor: make cursor rebasing cursor type-agnostic This commit refactors `cursor_rebase` into `cursor_update_image`, and moves sending pointer events to the two existing call sites. This will enable this code to be reused for tablets. Refs #5232
5 years ago
enum wlr_edges find_resize_edge(struct sway_container *cont,
Do not search for edges on subsurfaces, fix #4381 Subsurfaces (in most cases popups) aren't decorated by sway and will never have any borders, but may be drawn beyond container boundaries producing false positive when searching for edge. So we want to skip edge search when handling mouse event on subsurface.
5 years ago
struct wlr_surface *surface, struct sway_cursor *cursor) {
enum wlr_edges edge = find_edge(cont, surface, cursor);
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
if (edge && !container_is_floating(cont) && edge_is_external(cont, edge)) {
return WLR_EDGE_NONE;
}
return edge;
}
/**
* Return the mouse binding which matches modifier, click location, release,
* and pressed button state, otherwise return null.
*/
static struct sway_binding* get_active_mouse_binding(
struct seatop_default_event *e, list_t *bindings, uint32_t modifiers,
bool release, bool on_titlebar, bool on_border, bool on_content,
bool on_workspace, const char *identifier) {
uint32_t click_region =
((on_titlebar || on_workspace) ? BINDING_TITLEBAR : 0) |
((on_border || on_workspace) ? BINDING_BORDER : 0) |
((on_content || on_workspace) ? BINDING_CONTENTS : 0);
struct sway_binding *current = NULL;
for (int i = 0; i < bindings->length; ++i) {
struct sway_binding *binding = bindings->items[i];
if (modifiers ^ binding->modifiers ||
e->pressed_button_count != (size_t)binding->keys->length ||
release != (binding->flags & BINDING_RELEASE) ||
!(click_region & binding->flags) ||
(on_workspace &&
(click_region & binding->flags) != click_region) ||
(strcmp(binding->input, identifier) != 0 &&
strcmp(binding->input, "*") != 0)) {
continue;
}
bool match = true;
for (size_t j = 0; j < e->pressed_button_count; j++) {
uint32_t key = *(uint32_t *)binding->keys->items[j];
if (key != e->pressed_buttons[j]) {
match = false;
break;
}
}
if (!match) {
continue;
}
if (!current || strcmp(current->input, "*") == 0) {
current = binding;
if (strcmp(current->input, identifier) == 0) {
// If a binding is found for the exact input, quit searching
break;
}
}
}
return current;
}
/**
* Remove a button (and duplicates) from the sorted list of currently pressed
* buttons.
*/
static void state_erase_button(struct seatop_default_event *e,
uint32_t button) {
size_t j = 0;
for (size_t i = 0; i < e->pressed_button_count; ++i) {
if (i > j) {
e->pressed_buttons[j] = e->pressed_buttons[i];
}
if (e->pressed_buttons[i] != button) {
++j;
}
}
while (e->pressed_button_count > j) {
--e->pressed_button_count;
e->pressed_buttons[e->pressed_button_count] = 0;
}
}
/**
* Add a button to the sorted list of currently pressed buttons, if there
* is space.
*/
static void state_add_button(struct seatop_default_event *e, uint32_t button) {
if (e->pressed_button_count >= SWAY_CURSOR_PRESSED_BUTTONS_CAP) {
return;
}
size_t i = 0;
while (i < e->pressed_button_count && e->pressed_buttons[i] < button) {
++i;
}
size_t j = e->pressed_button_count;
while (j > i) {
e->pressed_buttons[j] = e->pressed_buttons[j - 1];
--j;
}
e->pressed_buttons[i] = button;
e->pressed_button_count++;
}
input/tablet: allow moving floating tablet v2 surfaces by pen input Refs #5293.
5 years ago
/*-------------------------------------------\
* Functions used by handle_tablet_tool_tip /
*-----------------------------------------*/
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
input/tablet: add seatop_down entry for tablet input Currently, when tablet input exits a window during an implicit grab, it passes focus to another window. For instance, this is problematic when trying to drag a scrollbar, and exiting the window &mdash; the scrollbar motion stops. Additionally, without `focus_follows_mouse no`, the tablet passes focus to whatever surface it goes over regardless of if there is an active implicit. If the tablet is over a surface that does not bind tablet handlers, sway will fall back to pointer emulation, and all of this works fine. It probably should have consistent behavior between emulated and not-emulated input, though. This commit adds a condition for entering seatop_down when a tablet's tool tip goes down, and exiting when it goes up. Since events won't be routed through seatop_default, this prevents windows losing focus during implicit grabs. Closes #5302.
5 years ago
static void handle_tablet_tool_tip(struct sway_seat *seat,
struct sway_tablet_tool *tool, uint32_t time_msec,
enum wlr_tablet_tool_tip_state state) {
input/cursor: refactor tablet tool tip events into seatops This commit moves tool tip event generation into seatops. In doing so, some corner cases where we'd erroneously (but likely harmlessly) generate both tablet and pointer events simultaneously are eliminated.
5 years ago
if (state == WLR_TABLET_TOOL_TIP_UP) {
wlr_tablet_v2_tablet_tool_notify_up(tool->tablet_v2_tool);
input/tablet: add seatop_down entry for tablet input Currently, when tablet input exits a window during an implicit grab, it passes focus to another window. For instance, this is problematic when trying to drag a scrollbar, and exiting the window &mdash; the scrollbar motion stops. Additionally, without `focus_follows_mouse no`, the tablet passes focus to whatever surface it goes over regardless of if there is an active implicit. If the tablet is over a surface that does not bind tablet handlers, sway will fall back to pointer emulation, and all of this works fine. It probably should have consistent behavior between emulated and not-emulated input, though. This commit adds a condition for entering seatop_down when a tablet's tool tip goes down, and exiting when it goes up. Since events won't be routed through seatop_default, this prevents windows losing focus during implicit grabs. Closes #5302.
5 years ago
return;
}
struct sway_cursor *cursor = seat->cursor;
struct wlr_surface *surface = NULL;
double sx, sy;
struct sway_node *node = node_at_coords(seat,
cursor->cursor->x, cursor->cursor->y, &surface, &sx, &sy);
input/tablet: allow moving floating tablet v2 surfaces by pen input Refs #5293.
5 years ago
if (!sway_assert(surface,
"Expected null-surface tablet input to route through pointer emulation")) {
return;
}
struct sway_container *cont = node && node->type == N_CONTAINER ?
node->sway_container : NULL;
input/tablet: focus appropriate node on tip down This commit makes tablet input more usable when `focus_follows_mouse` is set to `no`. Previously, tapping down on surfaces that bound tablet input would not switch focus, whereas tapping on surfaces that didn't (and hence went through pointer emulation) did.
5 years ago
if (wlr_surface_is_layer_surface(surface)) {
// Handle tapping a layer surface
struct wlr_layer_surface_v1 *layer =
wlr_layer_surface_v1_from_wlr_surface(surface);
if (layer->current.keyboard_interactive) {
seat_set_focus_layer(seat, layer);
}
} else if (cont) {
input/tablet: allow moving floating tablet v2 surfaces by pen input Refs #5293.
5 years ago
bool is_floating_or_child = container_is_floating_or_child(cont);
bool is_fullscreen_or_child = container_is_fullscreen_or_child(cont);
struct wlr_keyboard *keyboard = wlr_seat_get_keyboard(seat->wlr_seat);
bool mod_pressed = keyboard &&
(wlr_keyboard_get_modifiers(keyboard) & config->floating_mod);
// Handle beginning floating move
if (is_floating_or_child && !is_fullscreen_or_child && mod_pressed) {
seat_set_focus_container(seat,
seat_get_focus_inactive_view(seat, &cont->node));
seatop_begin_move_floating(seat, container_toplevel_ancestor(cont));
return;
}
input/tablet: allow moving tiling tablet v2 surfaces by pen input Closes #5293.
5 years ago
// Handle moving a tiling container
if (config->tiling_drag && mod_pressed && !is_floating_or_child &&
cont->fullscreen_mode == FULLSCREEN_NONE) {
seatop_begin_move_tiling(seat, cont);
return;
}
input/tablet: focus appropriate node on tip down This commit makes tablet input more usable when `focus_follows_mouse` is set to `no`. Previously, tapping down on surfaces that bound tablet input would not switch focus, whereas tapping on surfaces that didn't (and hence went through pointer emulation) did.
5 years ago
// Handle tapping on a container surface
seat_set_focus_container(seat, cont);
input/tablet: add seatop_down entry for tablet input Currently, when tablet input exits a window during an implicit grab, it passes focus to another window. For instance, this is problematic when trying to drag a scrollbar, and exiting the window &mdash; the scrollbar motion stops. Additionally, without `focus_follows_mouse no`, the tablet passes focus to whatever surface it goes over regardless of if there is an active implicit. If the tablet is over a surface that does not bind tablet handlers, sway will fall back to pointer emulation, and all of this works fine. It probably should have consistent behavior between emulated and not-emulated input, though. This commit adds a condition for entering seatop_down when a tablet's tool tip goes down, and exiting when it goes up. Since events won't be routed through seatop_default, this prevents windows losing focus during implicit grabs. Closes #5302.
5 years ago
seatop_begin_down(seat, node->sway_container, time_msec, sx, sy);
}
input/tablet: focus appropriate node on tip down This commit makes tablet input more usable when `focus_follows_mouse` is set to `no`. Previously, tapping down on surfaces that bound tablet input would not switch focus, whereas tapping on surfaces that didn't (and hence went through pointer emulation) did.
5 years ago
#if HAVE_XWAYLAND
// Handle tapping on an xwayland unmanaged view
else if (wlr_surface_is_xwayland_surface(surface)) {
struct wlr_xwayland_surface *xsurface =
wlr_xwayland_surface_from_wlr_surface(surface);
if (xsurface->override_redirect &&
wlr_xwayland_or_surface_wants_focus(xsurface)) {
struct wlr_xwayland *xwayland = server.xwayland.wlr_xwayland;
wlr_xwayland_set_seat(xwayland, seat->wlr_seat);
seat_set_focus_surface(seat, xsurface->surface, false);
}
}
#endif
input/tablet: allow moving floating tablet v2 surfaces by pen input Refs #5293.
5 years ago
wlr_tablet_v2_tablet_tool_notify_down(tool->tablet_v2_tool);
wlr_tablet_tool_v2_start_implicit_grab(tool->tablet_v2_tool);
input/tablet: add seatop_down entry for tablet input Currently, when tablet input exits a window during an implicit grab, it passes focus to another window. For instance, this is problematic when trying to drag a scrollbar, and exiting the window &mdash; the scrollbar motion stops. Additionally, without `focus_follows_mouse no`, the tablet passes focus to whatever surface it goes over regardless of if there is an active implicit. If the tablet is over a surface that does not bind tablet handlers, sway will fall back to pointer emulation, and all of this works fine. It probably should have consistent behavior between emulated and not-emulated input, though. This commit adds a condition for entering seatop_down when a tablet's tool tip goes down, and exiting when it goes up. Since events won't be routed through seatop_default, this prevents windows losing focus during implicit grabs. Closes #5302.
5 years ago
}
input/tablet: allow moving floating tablet v2 surfaces by pen input Refs #5293.
5 years ago
/*----------------------------------\
* Functions used by handle_button /
*--------------------------------*/
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
static void handle_button(struct sway_seat *seat, uint32_t time_msec,
struct wlr_input_device *device, uint32_t button,
enum wlr_button_state state) {
struct seatop_default_event *e = seat->seatop_data;
struct sway_cursor *cursor = seat->cursor;
// Determine what's under the cursor
struct wlr_surface *surface = NULL;
double sx, sy;
struct sway_node *node = node_at_coords(seat,
cursor->cursor->x, cursor->cursor->y, &surface, &sx, &sy);
Do not search for edges on subsurfaces, fix #4381 Subsurfaces (in most cases popups) aren't decorated by sway and will never have any borders, but may be drawn beyond container boundaries producing false positive when searching for edge. So we want to skip edge search when handling mouse event on subsurface.
5 years ago
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
struct sway_container *cont = node && node->type == N_CONTAINER ?
node->sway_container : NULL;
bool is_floating = cont && container_is_floating(cont);
bool is_floating_or_child = cont && container_is_floating_or_child(cont);
bool is_fullscreen_or_child = cont && container_is_fullscreen_or_child(cont);
Do not search for edges on subsurfaces, fix #4381 Subsurfaces (in most cases popups) aren't decorated by sway and will never have any borders, but may be drawn beyond container boundaries producing false positive when searching for edge. So we want to skip edge search when handling mouse event on subsurface.
5 years ago
enum wlr_edges edge = cont ? find_edge(cont, surface, cursor) : WLR_EDGE_NONE;
input: check pointer against nullptr
5 years ago
enum wlr_edges resize_edge = cont && edge ?
Do not search for edges on subsurfaces, fix #4381 Subsurfaces (in most cases popups) aren't decorated by sway and will never have any borders, but may be drawn beyond container boundaries producing false positive when searching for edge. So we want to skip edge search when handling mouse event on subsurface.
5 years ago
find_resize_edge(cont, surface, cursor) : WLR_EDGE_NONE;
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
bool on_border = edge != WLR_EDGE_NONE;
bool on_contents = cont && !on_border && surface;
bool on_workspace = node && node->type == N_WORKSPACE;
bool on_titlebar = cont && !on_border && !surface;
// Handle mouse bindings
struct wlr_keyboard *keyboard = wlr_seat_get_keyboard(seat->wlr_seat);
uint32_t modifiers = keyboard ? wlr_keyboard_get_modifiers(keyboard) : 0;
char *device_identifier = device ? input_device_get_identifier(device)
: strdup("*");
struct sway_binding *binding = NULL;
if (state == WLR_BUTTON_PRESSED) {
state_add_button(e, button);
binding = get_active_mouse_binding(e,
config->current_mode->mouse_bindings, modifiers, false,
on_titlebar, on_border, on_contents, on_workspace,
device_identifier);
} else {
binding = get_active_mouse_binding(e,
config->current_mode->mouse_bindings, modifiers, true,
on_titlebar, on_border, on_contents, on_workspace,
device_identifier);
state_erase_button(e, button);
}
free(device_identifier);
if (binding) {
seat_execute_command(seat, binding);
return;
}
// Handle clicking an empty workspace
if (node && node->type == N_WORKSPACE) {
input/seatop_default: release on empty workspace Instead of handling presses and releases on empty workspaces as setting focus to the workspace, handle releases by notifying the seat of a pointer action. This way DnDs are correctly released if the button is released over an empty workspace. This is achieved by removing the early return and letting the handle_button() call seat_pointer_notify_button() at the very end. Fixes #3932
5 years ago
if (state == WLR_BUTTON_PRESSED) {
seat_set_focus(seat, node);
}
seat_pointer_notify_button(seat, time_msec, button, state);
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
return;
}
// Handle clicking a layer surface
if (surface && wlr_surface_is_layer_surface(surface)) {
struct wlr_layer_surface_v1 *layer =
wlr_layer_surface_v1_from_wlr_surface(surface);
if (layer->current.keyboard_interactive) {
seat_set_focus_layer(seat, layer);
}
seat_pointer_notify_button(seat, time_msec, button, state);
return;
}
// Handle tiling resize via border
if (cont && resize_edge && button == BTN_LEFT &&
state == WLR_BUTTON_PRESSED && !is_floating) {
input: tweak resize behavior to not change tab focus on border click If a resize is triggered on a tabbed or stacked container, change focus to the tab which already had inactive focus, rather than to the tab whose border was clicked -- otherwise, we'd change the active tab when the user probably just wanted to resize.
5 years ago
// If a resize is triggered on a tabbed or stacked container, change
// focus to the tab which already had inactive focus -- otherwise, we'd
// change the active tab when the user probably just wanted to resize.
struct sway_container *cont_to_focus = cont;
enum sway_container_layout layout = container_parent_layout(cont);
if (layout == L_TABBED || layout == L_STACKED) {
cont_to_focus = seat_get_focus_inactive_view(seat, &cont->parent->node);
}
seat_set_focus_container(seat, cont_to_focus);
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
seatop_begin_resize_tiling(seat, cont, edge);
return;
}
// Handle tiling resize via mod
bool mod_pressed = keyboard &&
(wlr_keyboard_get_modifiers(keyboard) & config->floating_mod);
if (cont && !is_floating_or_child && mod_pressed &&
state == WLR_BUTTON_PRESSED) {
uint32_t btn_resize = config->floating_mod_inverse ?
BTN_LEFT : BTN_RIGHT;
if (button == btn_resize) {
edge = 0;
edge |= cursor->cursor->x > cont->x + cont->width / 2 ?
WLR_EDGE_RIGHT : WLR_EDGE_LEFT;
edge |= cursor->cursor->y > cont->y + cont->height / 2 ?
WLR_EDGE_BOTTOM : WLR_EDGE_TOP;
const char *image = NULL;
if (edge == (WLR_EDGE_LEFT | WLR_EDGE_TOP)) {
image = "nw-resize";
} else if (edge == (WLR_EDGE_TOP | WLR_EDGE_RIGHT)) {
image = "ne-resize";
} else if (edge == (WLR_EDGE_RIGHT | WLR_EDGE_BOTTOM)) {
image = "se-resize";
} else if (edge == (WLR_EDGE_BOTTOM | WLR_EDGE_LEFT)) {
image = "sw-resize";
}
cursor_set_image(seat->cursor, image, NULL);
seat_set_focus_container(seat, cont);
seatop_begin_resize_tiling(seat, cont, edge);
return;
}
}
// Handle beginning floating move
if (cont && is_floating_or_child && !is_fullscreen_or_child &&
state == WLR_BUTTON_PRESSED) {
uint32_t btn_move = config->floating_mod_inverse ? BTN_RIGHT : BTN_LEFT;
Remove redundant checks
5 years ago
if (button == btn_move && (mod_pressed || on_titlebar)) {
input/seatop_default: fix focusing floating titles When clicking on the titlebar of a floating container (or descendant of a floating container), the top-level floating container was being focused and then allowing you to move the top-level floating container. This made it so you couldn't switch to a different tab/stack within the floating container. With this patch, the focus inactive view for the container that the titlebar is associated with is focused, then the traversal to the top-level floating container is performed to use with the move floating operation.
5 years ago
seat_set_focus_container(seat,
seat_get_focus_inactive_view(seat, &cont->node));
tree/container: introduce `container_toplevel_ancestor` helper This allows us to not have to explicitly write the same while loop everywhere.
5 years ago
seatop_begin_move_floating(seat, container_toplevel_ancestor(cont));
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
return;
}
}
// Handle beginning floating resize
if (cont && is_floating_or_child && !is_fullscreen_or_child &&
state == WLR_BUTTON_PRESSED) {
// Via border
if (button == BTN_LEFT && resize_edge != WLR_EDGE_NONE) {
seatop_begin_resize_floating(seat, cont, resize_edge);
return;
}
// Via mod+click
uint32_t btn_resize = config->floating_mod_inverse ?
BTN_LEFT : BTN_RIGHT;
if (mod_pressed && button == btn_resize) {
tree/container: introduce `container_toplevel_ancestor` helper This allows us to not have to explicitly write the same while loop everywhere.
5 years ago
struct sway_container *floater = container_toplevel_ancestor(cont);
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
edge = 0;
edge |= cursor->cursor->x > floater->x + floater->width / 2 ?
WLR_EDGE_RIGHT : WLR_EDGE_LEFT;
edge |= cursor->cursor->y > floater->y + floater->height / 2 ?
WLR_EDGE_BOTTOM : WLR_EDGE_TOP;
seatop_begin_resize_floating(seat, floater, edge);
return;
}
}
// Handle moving a tiling container
if (config->tiling_drag && (mod_pressed || on_titlebar) &&
state == WLR_BUTTON_PRESSED && !is_floating_or_child &&
cont && cont->fullscreen_mode == FULLSCREEN_NONE) {
struct sway_container *focus = seat_get_focused_container(seat);
bool focused = focus == cont || container_has_ancestor(focus, cont);
if (on_titlebar && !focused) {
node = seat_get_focus_inactive(seat, &cont->node);
seat_set_focus(seat, node);
}
// If moving a container by it's title bar, use a threshold for the drag
if (!mod_pressed && config->tiling_drag_threshold > 0) {
seatop_begin_move_tiling_threshold(seat, cont);
} else {
seatop_begin_move_tiling(seat, cont);
}
return;
}
// Handle mousedown on a container surface
if (surface && cont && state == WLR_BUTTON_PRESSED) {
seat_set_focus_container(seat, cont);
seatop_begin_down(seat, cont, time_msec, sx, sy);
seat_pointer_notify_button(seat, time_msec, button, WLR_BUTTON_PRESSED);
return;
}
// Handle clicking a container surface or decorations
seatop_default: only focus container on press This matches i3's behavior of only focusing a container when pressed. This allows for `bindsym button1 nop`, `bindsym BTN_LEFT nop`, or `bindcode 272 nop` to be used to disable focusing when clicking on the title (or with additional flags to bind{code,sym} other portions of the container). Without this additional condition, the user would need both `bindsym button1 nop` and `bindsym --release button1 nop` to override both the pressed and released behavior.
5 years ago
if (cont && state == WLR_BUTTON_PRESSED) {
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
node = seat_get_focus_inactive(seat, &cont->node);
seat_set_focus(seat, node);
seat_pointer_notify_button(seat, time_msec, button, state);
return;
}
seatop_default: handle focus for unmanaged xwayland windows last Fixes #4707
5 years ago
#if HAVE_XWAYLAND
// Handle clicking on xwayland unmanaged view
if (surface && wlr_surface_is_xwayland_surface(surface)) {
struct wlr_xwayland_surface *xsurface =
wlr_xwayland_surface_from_wlr_surface(surface);
if (xsurface->override_redirect &&
wlr_xwayland_or_surface_wants_focus(xsurface)) {
struct wlr_xwayland *xwayland = server.xwayland.wlr_xwayland;
wlr_xwayland_set_seat(xwayland, seat->wlr_seat);
seat_set_focus_surface(seat, xsurface->surface, false);
seat_pointer_notify_button(seat, time_msec, button, state);
return;
}
}
#endif
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
seat_pointer_notify_button(seat, time_msec, button, state);
}
input: rename pointer handlers to be unambiguous This commit renames `motion` and `axis` handlers to `pointer_motion` and `pointer_axis`, respectively, to disambiguate them from their tablet (and future touch) handlers. `button` is left as-is, as it is generic across input devices.
5 years ago
/*------------------------------------------\
* Functions used by handle_pointer_motion /
*----------------------------------------*/
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
static void check_focus_follows_mouse(struct sway_seat *seat,
struct seatop_default_event *e, struct sway_node *hovered_node) {
struct sway_node *focus = seat_get_focus(seat);
// If a workspace node is hovered (eg. in the gap area), only set focus if
// the workspace is on a different output to the previous focus.
if (focus && hovered_node->type == N_WORKSPACE) {
struct sway_output *focused_output = node_get_output(focus);
struct sway_output *hovered_output = node_get_output(hovered_node);
if (hovered_output != focused_output) {
seat_set_focus(seat, seat_get_focus_inactive(seat, hovered_node));
}
return;
}
Clean up focus follows mouse logic Firstly, this fixes a recent regression where having `focus_follows_mouse yes` and hovering an inactive tab caused it to gain focus. The code was missing a view_is_visible check. The code is handling the logic for both focus_follows_mouse yes and focus_follows_mouse always, where the latter will apply when nudging the mouse after a workspace switch. However, the view_is_visible check didn't apply when using focus_follows_mouse always, so hovering a tab with that configuration would cause is to focus. This was a bug. When adding the view_is_visible check, it now applies to both yes and always. Note that the comment about the split container was wrong. At this point the hovered node cannot be a split container because it passed the node_is_view check. The comment has been removed. Lastly, the else condition is completely removed. This didn't appear to have any practical use. Setting focus to the result of seat_get_focus_inactive is very likely going to be a no op. There is a slim chance that this will break something, and if so I'd like to find out what so it can be properly documented in the code.
6 years ago
// This is where we handle the common case. We don't want to focus inactive
// tabs, hence the view_is_visible check.
if (node_is_view(hovered_node) &&
view_is_visible(hovered_node->sway_container->view)) {
// e->previous_node is the node which the cursor was over previously.
// If focus_follows_mouse is yes and the cursor got over the view due
// to, say, a workspace switch, we don't want to set the focus.
// But if focus_follows_mouse is "always", we do.
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
if (hovered_node != e->previous_node ||
config->focus_follows_mouse == FOLLOWS_ALWAYS) {
seat_set_focus(seat, hovered_node);
}
}
}
input: rename pointer handlers to be unambiguous This commit renames `motion` and `axis` handlers to `pointer_motion` and `pointer_axis`, respectively, to disambiguate them from their tablet (and future touch) handlers. `button` is left as-is, as it is generic across input devices.
5 years ago
static void handle_pointer_motion(struct sway_seat *seat, uint32_t time_msec,
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
double dx, double dy) {
struct seatop_default_event *e = seat->seatop_data;
struct sway_cursor *cursor = seat->cursor;
struct wlr_surface *surface = NULL;
double sx, sy;
struct sway_node *node = node_at_coords(seat,
cursor->cursor->x, cursor->cursor->y, &surface, &sx, &sy);
if (node && config->focus_follows_mouse != FOLLOWS_NO) {
check_focus_follows_mouse(seat, e, node);
}
input/cursor: make cursor rebasing cursor type-agnostic This commit refactors `cursor_rebase` into `cursor_update_image`, and moves sending pointer events to the two existing call sites. This will enable this code to be reused for tablets. Refs #5232
5 years ago
if (surface) {
if (seat_is_input_allowed(seat, surface)) {
wlr_seat_pointer_notify_enter(seat->wlr_seat, surface, sx, sy);
wlr_seat_pointer_notify_motion(seat->wlr_seat, time_msec, sx, sy);
}
} else {
cursor_update_image(cursor, node);
Use new wlroots API for clearing keyboard/pointer focus during grabs We are not allowed to do what we did in #5222 and pass a `NULL` surface wlr_seat_pointer_notify_enter(), and it's causing crashes when an xdg-shell popup is active (see #5294 and swaywm/wlroots#2161). Instead, solve #5220 using the new wlroots API introduced in swaywm/wlroots#2217.
5 years ago
wlr_seat_pointer_notify_clear_focus(seat->wlr_seat);
Don't send pointer motion when rebasing the cursor
6 years ago
}
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
struct sway_drag_icon *drag_icon;
wl_list_for_each(drag_icon, &root->drag_icons, link) {
if (drag_icon->seat == seat) {
drag_icon_update_position(drag_icon);
}
}
e->previous_node = node;
}
input: refactor tablet motion into seatop handler This commit moves tablet motion logic into a seatop handler. As a side-effect of seatop implementations being able to receive tablet motion events, fixes #5232.
5 years ago
static void handle_tablet_tool_motion(struct sway_seat *seat,
input/tablet: simplify parameter plumbing for tablet references This is a small cleanup commit for removing `sway_tablet` parameters from functions that already accept `sway_tablet_tool`, since the tablet reference can be accessed through `tool->tablet`.
5 years ago
struct sway_tablet_tool *tool, uint32_t time_msec, double dx, double dy) {
input: refactor tablet motion into seatop handler This commit moves tablet motion logic into a seatop handler. As a side-effect of seatop implementations being able to receive tablet motion events, fixes #5232.
5 years ago
struct seatop_default_event *e = seat->seatop_data;
struct sway_cursor *cursor = seat->cursor;
struct wlr_surface *surface = NULL;
double sx, sy;
struct sway_node *node = node_at_coords(seat,
cursor->cursor->x, cursor->cursor->y, &surface, &sx, &sy);
if (node && config->focus_follows_mouse != FOLLOWS_NO) {
check_focus_follows_mouse(seat, e, node);
}
if (surface) {
if (seat_is_input_allowed(seat, surface)) {
wlr_tablet_v2_tablet_tool_notify_proximity_in(tool->tablet_v2_tool,
input/tablet: simplify parameter plumbing for tablet references This is a small cleanup commit for removing `sway_tablet` parameters from functions that already accept `sway_tablet_tool`, since the tablet reference can be accessed through `tool->tablet`.
5 years ago
tool->tablet->tablet_v2, surface);
input: refactor tablet motion into seatop handler This commit moves tablet motion logic into a seatop handler. As a side-effect of seatop implementations being able to receive tablet motion events, fixes #5232.
5 years ago
wlr_tablet_v2_tablet_tool_notify_motion(tool->tablet_v2_tool, sx, sy);
}
} else {
cursor_update_image(cursor, node);
wlr_tablet_v2_tablet_tool_notify_proximity_out(tool->tablet_v2_tool);
}
struct sway_drag_icon *drag_icon;
wl_list_for_each(drag_icon, &root->drag_icons, link) {
if (drag_icon->seat == seat) {
drag_icon_update_position(drag_icon);
}
}
e->previous_node = node;
}
input: rename pointer handlers to be unambiguous This commit renames `motion` and `axis` handlers to `pointer_motion` and `pointer_axis`, respectively, to disambiguate them from their tablet (and future touch) handlers. `button` is left as-is, as it is generic across input devices.
5 years ago
/*----------------------------------------\
* Functions used by handle_pointer_axis /
*--------------------------------------*/
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
static uint32_t wl_axis_to_button(struct wlr_event_pointer_axis *event) {
switch (event->orientation) {
case WLR_AXIS_ORIENTATION_VERTICAL:
return event->delta < 0 ? SWAY_SCROLL_UP : SWAY_SCROLL_DOWN;
case WLR_AXIS_ORIENTATION_HORIZONTAL:
return event->delta < 0 ? SWAY_SCROLL_LEFT : SWAY_SCROLL_RIGHT;
default:
sway_log(SWAY_DEBUG, "Unknown axis orientation");
return 0;
}
}
input: rename pointer handlers to be unambiguous This commit renames `motion` and `axis` handlers to `pointer_motion` and `pointer_axis`, respectively, to disambiguate them from their tablet (and future touch) handlers. `button` is left as-is, as it is generic across input devices.
5 years ago
static void handle_pointer_axis(struct sway_seat *seat,
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
struct wlr_event_pointer_axis *event) {
struct sway_input_device *input_device =
event->device ? event->device->data : NULL;
struct input_config *ic =
input_device ? input_device_get_config(input_device) : NULL;
struct sway_cursor *cursor = seat->cursor;
struct seatop_default_event *e = seat->seatop_data;
// Determine what's under the cursor
struct wlr_surface *surface = NULL;
double sx, sy;
struct sway_node *node = node_at_coords(seat,
cursor->cursor->x, cursor->cursor->y, &surface, &sx, &sy);
struct sway_container *cont = node && node->type == N_CONTAINER ?
node->sway_container : NULL;
Do not search for edges on subsurfaces, fix #4381 Subsurfaces (in most cases popups) aren't decorated by sway and will never have any borders, but may be drawn beyond container boundaries producing false positive when searching for edge. So we want to skip edge search when handling mouse event on subsurface.
5 years ago
enum wlr_edges edge = cont ? find_edge(cont, surface, cursor) : WLR_EDGE_NONE;
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
bool on_border = edge != WLR_EDGE_NONE;
bool on_titlebar = cont && !on_border && !surface;
bool on_titlebar_border = cont && on_border &&
cursor->cursor->y < cont->content_y;
bool on_contents = cont && !on_border && surface;
bool on_workspace = node && node->type == N_WORKSPACE;
float scroll_factor =
(ic == NULL || ic->scroll_factor == FLT_MIN) ? 1.0f : ic->scroll_factor;
bool handled = false;
// Gather information needed for mouse bindings
struct wlr_keyboard *keyboard = wlr_seat_get_keyboard(seat->wlr_seat);
uint32_t modifiers = keyboard ? wlr_keyboard_get_modifiers(keyboard) : 0;
struct wlr_input_device *device =
input_device ? input_device->wlr_device : NULL;
char *dev_id = device ? input_device_get_identifier(device) : strdup("*");
uint32_t button = wl_axis_to_button(event);
// Handle mouse bindings - x11 mouse buttons 4-7 - press event
struct sway_binding *binding = NULL;
state_add_button(e, button);
binding = get_active_mouse_binding(e, config->current_mode->mouse_bindings,
modifiers, false, on_titlebar, on_border, on_contents, on_workspace,
dev_id);
if (binding) {
seat_execute_command(seat, binding);
handled = true;
}
// Scrolling on a tabbed or stacked title bar (handled as press event)
if (!handled && (on_titlebar || on_titlebar_border)) {
enum sway_container_layout layout = container_parent_layout(cont);
if (layout == L_TABBED || layout == L_STACKED) {
struct sway_node *tabcontainer = node_get_parent(node);
struct sway_node *active =
seat_get_active_tiling_child(seat, tabcontainer);
list_t *siblings = container_get_siblings(cont);
int desired = list_find(siblings, active->sway_container) +
round(scroll_factor * event->delta_discrete);
if (desired < 0) {
desired = 0;
} else if (desired >= siblings->length) {
desired = siblings->length - 1;
}
struct sway_node *old_focus = seat_get_focus(seat);
struct sway_container *new_sibling_con = siblings->items[desired];
struct sway_node *new_sibling = &new_sibling_con->node;
struct sway_node *new_focus =
seat_get_focus_inactive(seat, new_sibling);
if (node_has_ancestor(old_focus, tabcontainer)) {
seat_set_focus(seat, new_focus);
} else {
// Scrolling when focus is not in the tabbed container at all
seat_set_raw_focus(seat, new_sibling);
seat_set_raw_focus(seat, new_focus);
seat_set_raw_focus(seat, old_focus);
}
handled = true;
}
}
// Handle mouse bindings - x11 mouse buttons 4-7 - release event
binding = get_active_mouse_binding(e, config->current_mode->mouse_bindings,
modifiers, true, on_titlebar, on_border, on_contents, on_workspace,
dev_id);
state_erase_button(e, button);
if (binding) {
seat_execute_command(seat, binding);
handled = true;
}
free(dev_id);
if (!handled) {
wlr_seat_pointer_notify_axis(cursor->seat->wlr_seat, event->time_msec,
event->orientation, scroll_factor * event->delta,
round(scroll_factor * event->delta_discrete), event->source);
}
}
/*----------------------------------\
* Functions used by handle_rebase /
*--------------------------------*/
static void handle_rebase(struct sway_seat *seat, uint32_t time_msec) {
struct seatop_default_event *e = seat->seatop_data;
struct sway_cursor *cursor = seat->cursor;
struct wlr_surface *surface = NULL;
double sx = 0.0, sy = 0.0;
e->previous_node = node_at_coords(seat,
cursor->cursor->x, cursor->cursor->y, &surface, &sx, &sy);
input/cursor: make cursor rebasing cursor type-agnostic This commit refactors `cursor_rebase` into `cursor_update_image`, and moves sending pointer events to the two existing call sites. This will enable this code to be reused for tablets. Refs #5232
5 years ago
if (surface) {
if (seat_is_input_allowed(seat, surface)) {
wlr_seat_pointer_notify_enter(seat->wlr_seat, surface, sx, sy);
input/pointer: send pointer enter event on confine warp The spec has this to say about sending events on confine creation: Whenever the confinement is activated, it is guaranteed that the surface the pointer is confined to will already have received pointer focus and that the pointer will be within the region passed to the request creating this object. ...and on region update: If warped, a wl_pointer.motion event will be emitted, but no wp_relative_pointer.relative_motion event. Prior to this patch, sway did neither, and updated the hardware cursor position without notifying the underlying surface until the next motion event. This led to inconsistent results, especially in applications that draw their own software cursor.
5 years ago
wlr_seat_pointer_notify_motion(seat->wlr_seat, time_msec, sx, sy);
input/cursor: make cursor rebasing cursor type-agnostic This commit refactors `cursor_rebase` into `cursor_update_image`, and moves sending pointer events to the two existing call sites. This will enable this code to be reused for tablets. Refs #5232
5 years ago
}
} else {
cursor_update_image(cursor, e->previous_node);
Use new wlroots API for clearing keyboard/pointer focus during grabs We are not allowed to do what we did in #5222 and pass a `NULL` surface wlr_seat_pointer_notify_enter(), and it's causing crashes when an xdg-shell popup is active (see #5294 and swaywm/wlroots#2161). Instead, solve #5220 using the new wlroots API introduced in swaywm/wlroots#2217.
5 years ago
wlr_seat_pointer_notify_clear_focus(seat->wlr_seat);
input/cursor: make cursor rebasing cursor type-agnostic This commit refactors `cursor_rebase` into `cursor_update_image`, and moves sending pointer events to the two existing call sites. This will enable this code to be reused for tablets. Refs #5232
5 years ago
}
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
}
static const struct sway_seatop_impl seatop_impl = {
.button = handle_button,
input: rename pointer handlers to be unambiguous This commit renames `motion` and `axis` handlers to `pointer_motion` and `pointer_axis`, respectively, to disambiguate them from their tablet (and future touch) handlers. `button` is left as-is, as it is generic across input devices.
5 years ago
.pointer_motion = handle_pointer_motion,
.pointer_axis = handle_pointer_axis,
input/tablet: add seatop_down entry for tablet input Currently, when tablet input exits a window during an implicit grab, it passes focus to another window. For instance, this is problematic when trying to drag a scrollbar, and exiting the window &mdash; the scrollbar motion stops. Additionally, without `focus_follows_mouse no`, the tablet passes focus to whatever surface it goes over regardless of if there is an active implicit. If the tablet is over a surface that does not bind tablet handlers, sway will fall back to pointer emulation, and all of this works fine. It probably should have consistent behavior between emulated and not-emulated input, though. This commit adds a condition for entering seatop_down when a tablet's tool tip goes down, and exiting when it goes up. Since events won't be routed through seatop_default, this prevents windows losing focus during implicit grabs. Closes #5302.
5 years ago
.tablet_tool_tip = handle_tablet_tool_tip,
input: refactor tablet motion into seatop handler This commit moves tablet motion logic into a seatop handler. As a side-effect of seatop implementations being able to receive tablet motion events, fixes #5232.
5 years ago
.tablet_tool_motion = handle_tablet_tool_motion,
Introduce default seatop This introduces a `default` seat operation which is used when no mouse buttons are being held. This means there is now always a seat operation in progress. It allows us to separate `default` code from the standard cursor management code. The sway_seatop_impl struct has gained callbacks `axis`, `rebase` and `end`, and lost callbacks `finish` and `abort`. `axis` and `rebase` are only used by the default seatop. `end` is called when a seatop is being replaced by another one and allows the seatop to free any resources, though no seatop currently needs to do this. `finish` is no longer required, as each seatop can gracefully finish in their `button` callback. And `abort` is not needed, as calling `end` would achieve the same thing. The struct has also gained a bool named allow_set_cursor which allows the client to set a new cursor during `default` and `down` seatops. Seatops would previously store which button they were started with and stop when that button was released. This behaviour is changed so that it only ends once all buttons are released. So you can start a drag with $mod+left, then click and hold right, release left and it'll continue dragging while the right button is held. The motion callback now accepts dx and dy. Most seatops don't use this as they store the cursor position when the seatop is started and compare it with the current cursor position. This approach doesn't make sense for the default seatop though, hence why dx and dy are needed. The pressed_buttons array has been moved from the sway_cursor struct to the default seatop's data. This is only used for the default seatop to check bindings. The total pressed button count remains in the sway_cursor struct though, because all the other seatops check it to know if they should end. The `down` seatop no longer has a `moved` property. This was used to track if the cursor moved and to recheck focus_follows_mouse, but seems to work without it. The logic for focus_follows_mouse has been refactored. As part of this I've removed the call to wlr_seat_keyboard_has_grab as we don't appear to use keyboard grabs. The functions for handling relative motion, absolute motion and tool axis have been changed. Previously the handler functions were handle_cursor_motion, handle_cursor_motion_absolute and handle_tool_axis. The latter two both called cursor_motion_absolute. Both handle_cursor_motion and cursor_motion_absolute did very similar things. These are now simplified into three handlers and a single common function called cursor_motion. All three handlers call cursor_motion. As cursor_motion works with relative distances, the absolute and tool axis handlers convert them to relative first.
6 years ago
.rebase = handle_rebase,
.allow_set_cursor = true,
};
void seatop_begin_default(struct sway_seat *seat) {
seatop_end(seat);
struct seatop_default_event *e =
calloc(1, sizeof(struct seatop_default_event));
sway_assert(e, "Unable to allocate seatop_default_event");
seat->seatop_impl = &seatop_impl;
seat->seatop_data = e;
seatop_rebase(seat, 0);
}