We used to have a list of X errors we should ignore in case they do
occur. This commit expands that functionality to also allow us aborting
on certain errors.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
We set event sequence number to the last sequence xlib knows about to
silence its complaint about missing sequence numbers, but we forgot to
restore it back afterwards.
This used to break error ignoring mechanism in `should_ignore`. In the
last commit we updated it to use full_sequence which incidently fixed
this problem. But let's restore the sequence number anyway for good
measure.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Changes to frame extents were only tracked if the frame was visible, but
we have to keep this information current even for invisible frames to
(not) render the correct area.
When a window is moved during fade-out, configure_win won't add the old
window extent to damage (because it's unmapping), but the new window
extent will be added to damage when the window is rendered. And the
window will be rendered in its new location. Thus the damage for the old
window extent is missing.
We could fix this by adding damage in configure_win for unmapping window
too. But in general we don't want to move a window while it's fading
out. So we shadow the window geometry. The shadow geometry is updated in
configure_win, and the update will only propagate to real geometry in
win_process_update_flags.
This also fix the case where a window is unmapped, moved, then mapped
all in the same render cycle. map_win_start is supposed to add the old
window extent to damage in that case, but the old window extent is
already overwritten when we reach map_win_start. The shadow geometry
fixes that.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Make unmapped window events work mostly like a mapped window, except
flags set on unmapped windows aren't processed until the window is
mapped.
Hopefully this unifies some of the code paths and reduce corner cases.
Should fix#525
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
If a window receives multiple configure notifies in between 2 frames, we
add all the in between positions into damage, when we really just need
the starting and the end position.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Previously, when some explicitly checked property (e.g.
_NET_WM_WINDOW_OPACITY) changed, it won't trigger a
win_on_factor_change, and the rules will not be re-evaluated. Because
that property stale flag will be cleared after we explicitly check the
property, so it won't be detected as a tracked property change.
Here, we instead set WIN_FLAGS_FACTOR_CHANGED directly when a tracked
property changed.
Fixes: f3ff7eff8c
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Part of the configure notify handling which requires querying the X
server, has been moved into the X critical section.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
* src/event.c: expose_root: region is not freed
* backend/xrender/xrender.c:
* bind_pixmap: the reply `r` is not freed
* blur: reg_op_resized is not freed
Spotted-by: Samuel Hand <samuel.d.hand@gmail.com>
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
This is a workaround for #301. This doesn't fix the bug, but by
allocating X resources much less frequently, this should make the
bug almost never happen.
Also, it might generally be a good idea to not create/destroy X
resources so often. (XFixes Region accounts for >99% of the resource
creations/destructions)
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Mark it CLIENT_STALE and clear the client instead of recheck it
immediately. It's better to recheck when we have the server grabbed.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Make sure the window struct is indeed destroyed and freed when a window
is reparented away from root. So we won't find it later and get
confused.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Don't just mark it as CLIENT_STALE. To make sure we don't later find the
old frame window as the frame of the client.
Fix assertion failure in ev_destroy_notify.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
This reverts commit 04fe4a76b2.
This brings back the previous incomplete fix attempt for #299. See the
commit message of the revert for why it's incomplete.
A different fix is then attempted, see commit xxxxxxx for how that fix
works.
However, the second fix is incomplete by itself as well. The reason is
that i3 reparent the real window to the frame first, before destroying
the placeholder client of that frame. So briefly, that frame would have
2 client windows. And the frame is mapped before the placeholder is
destroyed. So even though we only call win_recheck_client when/if the
frame window is mapped, it can still be called when there are 2 client
windows, it would pick up the wrong client window in that case.
So what we need is to combine both fixes.
The second fix makes sure we are up to date on the client window
information when we starts to listen for events on the frame window;
while the first fix would keep us up to date afterwards.
This revert also includes a fix for assertion failure raised in #371
See #299 for root cause of the bug.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Instead of handling reparent notify on the spot by updating the client
windows, setup a flag on the window and call win_recheck_client later.
This makes handling of complex scenarios easier. As example, see the
case in issue #299.
Note this is not a complete fix for #299
* Add new field `opacity_target_old` to `struct managed_win` for
tracking relevant `opacity_target` changes.
* Smoothly fade blur-texture opacity on window opacity changes (based on
window opacity), when the window was or will be fully transparent
(`w->opacity ~< 0.004`).
* Fixed alpha-clipping of the blur-texture when using `inactive-opacity` or
repeatedly setting window opacity with large fade intervals (should fix#314).
These two flags are intended for subtly different things. I can probably
justify the distinction, but it's definitely difficult to explain. And
there is no obvious benefits to keep them separate.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Commits reverted:
b652e8b58dbdf809d039e9ab970989481ac54f67
While those commits themselves could be useful even if they don't
completely fix#299, they come with the risks of introduce more
unforeseen bugs, which outweigh their benefit, so revert them.
A brief explanation of their problem:
The fix hinges on the destroy event of the client window to work.
However, the client window could be destroyed so fast, before we even
handle the map, or even the create, event of the frame. And we won't be
listening for substructure events on the frame at the point the client
window is destroyed. Thus completely miss the client window destroy
event.
Because of the inherent racy nature of Xorg, this approach is really
difficult to make work.
Fixes#371
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
We detach the client window from its frame when it's destroyed, so later
a new window can be attached to that frame. But turns out that's not
enough.
When i3 restarts, it attaches a placeholder window to the frame, then
the real window, and only after that, it will destroy the placeholder.
The real window will fail to attach as the frame already has a client.
When the placeholder is destroyed, we have to call recheck client to
pick up the real window that failed to attach previously.
Related: #299
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Basically, don't ignore the reparent notify if the window is already a
client window. We already unmark it from its previous frame, so this
commit make sure it gets marked with its new frame.
Related: #299
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
We do nothing when a non-wm-frame window is destroyed. This will cause
trouble if a wm-frame window is reused (i.e. its child is destroyed then
a new child is reparented to it), because we didn't clear client_win.
So this commit adds a call to win_unmark_client for that case.
Related: #299
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Previously, root ConfigureNotify is handled immediately, by resetting
the backend, which in turn releases all the window images. This puts all
the windows into a state where they don't have images attached, which
they really should be in when the screen is redirected.
(To expand a little, a window without images should only exist if:
* It's an unmanaged window.
* Screen is unredirected.)
Normally, this kind of window could be fine, as the next render phase
will re-acquire images for them. However, if a window in this state is
destroyed with fading enabled, then the render phase won't try to
acquire images for it, causing it to go into the main rendering function
without images attached, and trigger an assertion.
This commit delays the handling of root ConfigureNotify until the render
phase. This way, the images will be immediately re-acquired after they
are released, thus prevent this problem from happening.
Also adds a testcase for this.
Fixes#357
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Unredirect the screen when some window sets _NET_WM_BYPASS_COMPOSITOR
to 1, but ignore the case where some window requests the screen to be
redirected by setting _NET_WM_BYPASS_COMPOSITOR to 2.
Closes#267
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
leftovers:
1) config file path. Has to implement compatibility functionalities before
we can change it.
2) links in man pages. Has to migrate the repo first.
3) _COMPTON_SHADOW, it has become a defacto standard, so we have to keep
supporting it.
4) dbus names, undecided whether we should/could change it.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
A quick search on GitHub showing the _COMPTON_SHADOW is actively being
used (most noteably, by polybar). It is not a difficult thing to support,
and it doesn't have that much runtime overhead. So remove the option
that guards this feature, make it enabled by default.
(Side note, I saw lots of people putting this option in their config
files, but this is actually a command line only option.)
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
Firstly, this breaks the assumption that when we redirect the screen,
all window is either mapped or unmapped, causing a assertion failure.
Secondly, if transition window into fading when the screen is not
redirected, the fading will not progress until the screen is redirected,
probably not the desired result.
Fixes#246
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
win_get_name _updates_ the name of the window by querying the X server,
thus potentially changes the behaviour of the compositor when logging is
turned on (e.g. we could fail to detect window name changes, because the
name was blindly updated in a logging call without handling the name change).
We shouldn't do that.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
And a window update flag for mapping the window.
Also make sure related functions consider the case where the given window
has pending updates.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
rename map_win, unmap_win, destroy_win to map_win_start,
unmap_win_start, destroy_win_start respectively. To clarify their
intended functions. Also rename the corresponding finish_* functions to
*_finish so they are consistent.
Also some very minor code clean ups.
Signed-off-by: Yuxuan Shui <yshuiv7@gmail.com>
