This commit rewrites the nested action parser into append_actions() which
is used by following commits. At this point, it's not used yet and parsing
"If" action is temporarily disabled.
For example, the following node:
<keybind name.action="ShowMenu" menu.action="root-menu"
x.position.action="1" y.position.action="2" />
is converted to:
<keybind>
<action>
<name>ShowMenu</name>
<menu>root-menu</menu>
<position>
<x>1</x>
<y>2</y>
</position>
</action>
</keybind>
...before processing the entire xml tree. This is a preparation to prevent
breaking changes when we refactor rcxml.c to use recursion instead of
encoding nodes into dotted strings.
This is a common practice in C projects, which simply enforces that
each header must compile cleanly without implicit dependencies on
other headers (see also the previous commit).
The lifetime of the "current_" variables (current_menu, current_item,
current_item_action) is very difficult to understand from reading the
code. It appears that e.g. current_menu could still point to a previous
menu when starting to parse a new one, with unpredictable results.
Let's use a context struct when parsing, and consistently initialize
it when beginning to build a new menu.
Lightly tested with:
- default menus (no menu.xml)
- example static menu from labwc.github.io/getting-started.html
- an added "client-list-combined-menu" sub-menu
- pipe menu generated by `labwc-menu-generator -p`
v2: style fix
Hi-res mice produces mulitple scroll events with `delta_discrete` != 0
during a single "click". This patch makes them trigger `Scroll` actions
only when the accumulated `delta_discrete` exceeds 120 (= 1 click).
See https://lists.freedesktop.org/archives/wayland-devel/2019-April/040377.html
for how hi-res scroll events are reported.
Having multiple declarations of the same static data (where one is
considered "tentative") is kind of an obscure C feature -- I didn't
even know the name of it until today. It's also forbidden in C++.
In the case of circular dependencies between static data <-> function,
the more typical pattern is to forward-declare the function, then the
data, then provide the function definition. Let's follow that pattern.
'restrict' is harmful as it encourages the compiler to make dangerous
assumptions while increasing cognitive load on the human programmer.
The extra 1% (or whatever) of performance here is not worth the cost.
img_svg_render() calls g_object_unref() on the RsvgHandle in its
error path, but the handle is owned by the shared lab_img_data
struct and will be double-freed later by lab_img_destroy().
The double-free was introduced when img_svg_load() was split from
img_svg_render(). The g_object_unref() should have been removed from
img_svg_render() but was missed.
Fixes: 16dbdc64e5
("ssd: rework titlebar button rendering")
Wlroots now destroys the wlr_output when granting a lease.
So we can't iterate through the outputs in the request after
granting the lease. This is also not necessary anymore because
they are already destroyed and thus removed from the layout.
This revises the changes done in 22d319c:
- Cancel defaulting to <dragLock>sticky<dragLock>. So labwc now disables
drag-lock by default, as libinput does.
- Update the options from [yes|no|sticky] to [timeout|no|yes] to
emphasize that the "sticky" mode (now "yes") is recommended when using
drag-lock.
Call seat_set_focus_layer(seat, NULL) in node-destroy-handler to avoid
seat->focused_layer becoming invalid and causing UAF issues in certain
situations like when outputs (and therefore layer-trees) are destroyed.
Fixes: #2863
Helped-by: @Consolatis
Alternative to 7bf08af which was reverted in the previous commit.
7bf08af fixed the problem that layer-shell clients are terminated when
it's unmapped, by sending configure events in node-destroy handler
rather than in unmap handler. But it caused a UAF bug when an output
with layer-shell clients is destroyed.
So this patch fixes the original issue by simply skipping the surface in
arrange_one_layer() if it's being unmapped.
Due to wlroots/pixman bug, the widened 1px buffer for the gradient
titlebar become translucent with WLR_RENDERER=pixman and
WLR_SCALE_FILTER_BILINEAR filter mode. This patch works around this
problem by using WLR_SCALE_FILTER_NEAREST filter mode if pixman
renderer is used.
The workarounds added in #2498 and #2437 fixed stuck key/modifier bug
caused by wlroots commit e218990. But now that the commit was reverted in
0.19, the workarounds are no longer needed.
Removing the workarounds also fixes a minor regression with Fcitx5+Firefox
that pressing Ctrl+Enter in an input box causes stuck modifier.
Only Vertical and SplitVertical gradients are supported,
and only for window.*.title.bg.
Not supported at this time:
- horizontal or diagonal gradients
- gradients for window.*.label.bg, buttons, or menus
- any type of border (raised, sunken, etc.)
