Sway doesn't update window visibility when a container changes

[WhyNotHugo]

Sway Version

sway version 1.10-dev-40ca4150b (Jun 10 2024, branch 'master')

I've seen this bug for several weeks now, I think it was introduced with the scene graph changes.

Debug Log

https://paste.sr.ht/~whynothugo/a7173261a6e1d238ddb739a040b70080921fe84b

Configuration File

exec foot
exec foot
# You'll need these mapping to easily reproduce the issue.
bindsym Mod4+a focus parent
bindsym Mod4+shift+s move container to scratchpad

Stack Trace

No crash

Description

• Start on an empty workspace
• Open a terminal
• Open another terminal
• focus parent
• move container to scratchpad
• BUG: Both windows are still there
• Switch to another workspace and back
• Windows are gone (they're now in the scratchpad).

Note that you can still interact with the windows in shrödinger's scratchpad until you switch to another workspace (when they disappear).

[WhyNotHugo]

There is another sequence of step with similar results, and I suspect they are the same issue:

• Open FOUR terminals.
• Switch to a tabbed layout.
• Focus the second terminal
• Split vertically
• Focus the fourth terminal
• move left twice
• The tab-bar shows the second tab as active, but the third tab is visible (although we never unfocused the terminal that is not part of the split inside the second tab).
• move left again (nothing happens on screen, although the window moves to the left inside the split)
• move left again. The focused terminal (originally on the far right) has now escaped the split without it ever having been rendered inside the split.

[WhyNotHugo]

I bisected the issue, but had to skip a lot of commits that either don't build, or render a black screen.

The first bad commit could be any of:

188811f
5b8b505
6d7b132
946fc80
08c484f
869baff
b38ed8b
bac3ab5
6e5fc4c
9a57966
0639bde
ed2724b
06ad734
c640c30
1e018e7
9c17cba
0e1a02b
5f0801b
2e53de8

I seems that the bug was introduced with the scene graph changes. I'm not sure that this information is of any use.

[Nefsen402]

I didn't know this issue existed. I'll look at it now.

[Nefsen402]

There is another sequence of step with similar results, and I suspect they are the same issue:

I cannot reproduce this issue. Mind sending a recording of the steps being reproduced?

[WhyNotHugo]

Sorry, I couldn't make a clear video showing the issue (you can't really understand what's going on in a video). I've made a reproduction script which should help you visualise the issue: #8292