A layer-shell surface can be unmapped if wlr_layer_shell_v1 is
destroyed or the client has committed a NULL buffer. Let's use the
previously introduced wlr_surface.unmap_commit to handle the latter
case instead; this is more consistent with the xdg_surface
implementation logic, where using the hook is more trouble than it's
worth.
Additionally, this commit adds an unconditional surface reset on
destroy, so popups are properly cleaned up even if originally created
with an unmapped layer-shell surface as a parent. Doing so with the
role unmap hook would either result in possibly resetting the surface
twice, which is suboptimal, or having an awkward
`if (mapped) { unmap() } else { reset() }` check.
This flag can be used to figure out whether a particular commit has
unmapped the surface. Private state for now in case we find a better
way to track this.
Client examples have been moved to another repo, but it seems I
forgot to delete some files.
Fixes: 0bb445eeff ("examples: split clients in separate repository")
Translating the right/bottom coordinates from offsets to absolute
coordinates in wlroots (rather than in the compositor) was supposed to
be more reliable, since wlroots had access to the X11 screen size.
It ended up being less reliable, because the screen size values
(xwm->screen->width_in_pixels/height_in_pixels) are not updated when the
output layout changes.
So let's remove the translation from wlroots, and let the compositor
figure it out. From what I can understand of the current XWayland code,
the X11 screen size should generally match the overall wlr_output_layout
bounding box, which the compositor has access to.
The client examples are useful to try out protocols, however they
don't need to live in the wlroots repository. Having both clients
and compositors in the same place is confusing. The wlroots API
changes often but protocols are set in stone.
The variable is named "libdrm" but it's a partial dependency with
just the headers. Reflect this in the name to avoid confusion (Meson
variables are global to the whole project).
We currently only perform non-blocking commits for non-modeset commits
with a buffer attached.
Perform non-blocking commits whenever there is no pending pageflip
event. If a non-blocking modeset commit fails, which can happen if the
driver implicitly added more CRTCs to the commit that we did not know we
had to wait for, retry with a blocking commit.
References: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/2239
The name "allow_artifacts" and associated description is very vague, and
theoretically allow for tearing behavior. Clarify that we only intend to
mean artifacts related to output configuration (e.g., modesets).
References: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3740
There were a couple of problems with this:
1. The behavior is unexpected. Typically objects in wlroots won't
also destroy objects that they depend on. For instance, wlr_scene_output
will not destroy the wlr_output when it's destroyed. It shouldn't be any
different for scene layouts.
2. This fixes a crash where because wlr_output_layout and wlr_scene_output
are both addons to wlr_output, we might get into a situation where
wl_list_for_each_safe might malfunction. See [1]
This means that the compositor needs to manually destroy the output
when they destroy the layout, hence ~breaking. Compositors can just
call `wlr_scene_output_destroy()` if they want to destroy both at the
same time.
[1] https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/4358#note_2106260
This reverts commit 1a731596c5.
Co-authored-by: Kirill Primak <vyivel@eclair.cafe>
It makes little sense to have a catch-all grab vaildation function,
considering that e.g. tablet tool implicit grabs are possible as well.
Besides, the function has always returned true anyway.
Instead of sending one request, waiting for the reply, and
repeating for all properties we're interested in, we can send all
property requests in one go and then wait for the server to reply.
The new struct rect_union is designed to make it easier to
efficiently accumulate a list of rectangles, and then operate
on an exact cover of their union.
Using rect_union, the times needed to added t rectangles, and then
compute their exact cover will be O(t), and something between Ω(t) and
O(t^2), depending on the rectangle arrangement. If one tries to do
the same by storing a pixman_region32_t and updating it with
pixman_region32_union_rect(), then total time needed would be between
Ω(t^2) and O(t^3), depending on the input. Without changing the public
API (data structure + rectangle ordering rules) for pixman_region32_t,
it is impossible to improve its worst case time.
