[Wayland-bugs] [Bug 88857] subsurface protocol is inconsistent regarding immediate commit vs deferred commit

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=88857

--- Comment #2 from Jonas Ådahl <jadahl at gmail.com> ---
Ah, right, the position should be considered part of the parents surface, true.
The issue that exists is though that the protocol explicitly says "on
wl_surface.commit", which is not necessarily the same time that the parent
surface gets its content updated. AFAIU the content of a surface is applied
either on

 - wl_surface.commit,
 - wl_subsurface.set_desync, or
 - when parent surface state is applied

So if we have this "tree", with one toplevel surface with a single subsurface
which itself has one single subsurface:

     T
     |
     s1
     |
     s2

If the following flow happens:

 1. s1::wl_subsurface.set_sync
 2. s1::wl_surface.attach
 3. s1::wl_surface.commit
 4. s2::wl_subsurface.set_desync
 5. s2::wl_subsurface.set_position
 6. s2::wl_surface.attach
 7. s2::wl_surface.commit
 8. s1::wl_subsurface.set_desync
 9. s1::wl_surface.attach
10. s1::wl_surface.commit

According to the spec the contents of s1 and s2 would be applied at 8, but the
position would be applied at 10, but if I understand you correctly, the
intention is that both the position and both s1's and s2's content should be
applied at 8.

It can be observed by running the test case in
https://bugzilla.gnome.org/show_bug.cgi?id=731494
on weston and mutter, and observe that they behave differently because of this
detail.

It could maybe be fixed by changing the protocol to no longer require an
explicit parent commit, but just an "parent content applied", I think, but can
we make such changes now?

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/wayland-bugs/attachments/20150203/508a19c3/attachment.html>