Scope popup.css to menupopup and panel custom elements
Categories
(Toolkit :: Themes, task, P3)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox77 | --- | affected |
People
(Reporter: ntim, Unassigned)
References
(Depends on 1 open bug, Blocks 1 open bug)
Details
Attachments
(1 file)
This is actually surprisingly straightforward, given that both these CEs already use shadow DOM and most (if not all) external styling already uses shadow parts.
Updated•5 years ago
|
| Reporter | ||
Comment 1•5 years ago
|
||
Updated•5 years ago
|
Comment 3•5 years ago
|
||
Backed out changeset 64f788053239 (bug 1625720) for mochitest failures in tests/chrome/test_panel.xhtml. CLOSED TREE
Log:
https://treeherder.mozilla.org/logviewer.html#/jobs?job_id=298846268&repo=autoland&lineNumber=11123
Push with failures:
https://treeherder.mozilla.org/#/jobs?repo=autoland&group_state=expanded&selectedJob=298847809&revision=64f788053239b1c3faa83e9f2340b1b1f82c6851
Backout:
https://hg.mozilla.org/integration/autoland/rev/023ec9bf4bd6e760d937071dbf8e716cf7219caf
| Reporter | ||
Comment 4•5 years ago
|
||
| Reporter | ||
Comment 5•5 years ago
|
||
https://treeherder.mozilla.org/#/jobs?repo=try&revision=91713493297963d8828f7911770ab2fd6c7dba41
It's not ideal, but this patch fixes the failure: https://hg.mozilla.org/try/rev/2b02c8ee22a6a1fc204a46e01f81a8b866490c82
I can't really tell if the failure was caused by the styling being applied too late (due to being loaded by the shadow root), or whether it was caused by the sole presence of the shadow DOM stylesheet in the panel custom elements. I'm guessing it's the former.
Either way, only the :host CSS rule is relevant to panels. The rest is of the popup.css stylesheet is only relevant to menupopup.
Dão, would you be ok with landing the patch I've linked above (or something similar) ?
Comment 6•5 years ago
|
||
Is the failure something that would affect real world users, or can we fix the test? Which of the properties you moved to xul.css is responsible for the failure? Can we move only that property instead of all?
| Reporter | ||
Comment 7•5 years ago
|
||
(In reply to Dão Gottwald [::dao] from comment #6)
Is the failure something that would affect real world users, or can we fix the test?
I honestly don't know.
According to the logs, it seems to crash around: https://searchfox.org/mozilla-central/rev/41c3ea3ee8eab9ce7b82932257cb80b703cbba67/toolkit/content/tests/chrome/window_panel.xhtml#71-73
for the "noautohide panel with titlebar" test
Which of the properties you moved to xul.css is responsible for the failure? Can we move only that property instead of all?
This is a Windows-only failure so I'm guessing only moving over the Windows properties would work. The only thing I know is that the min-width property is not needed as comment 4 shows.
I don't have a Windows machine and my time is limited until the end of May, so I really don't have time to figure out which exact property is behind it.
Dão, would you be ok with only moving over the Windows properties except for min-width, just to fix the regressions ? I can file a follow up bug and needinfo Neil Deakin on it for the time being.
Comment 8•5 years ago
|
||
(In reply to Tim Nguyen :ntim from comment #7)
Which of the properties you moved to xul.css is responsible for the failure? Can we move only that property instead of all?
This is a Windows-only failure so I'm guessing only moving over the Windows properties would work. The only thing I know is that the min-width property is not needed as comment 4 shows.
Can you narrow it down further to one property? Shouldn't be a lot of effort with a series of parallel try pushes...
| Reporter | ||
Comment 9•5 years ago
|
||
Comment 10•5 years ago
|
||
Comment 11•5 years ago
•
|
||
Backed out changeset 9432b1bd3d98 for causing crashes in LayerComposite.
Backout link: https://hg.mozilla.org/integration/autoland/rev/de714602be8edf8e239d4f89e8d13e59129b8132
Failure logs:
Comment 12•5 years ago
|
||
| Reporter | ||
Updated•5 years ago
|
Comment 13•5 years ago
|
||
| bugherder | ||
Comment 14•5 years ago
|
||
(In reply to Tim Nguyen :ntim from comment #7)
Dão, would you be ok with only moving over the Windows properties except for min-width, just to fix the regressions ? I can file a follow up bug and needinfo Neil Deakin on it for the time being.
The ifdef should have been around the whole rule, and the comment should have referenced the bug you were going to file. Can you please do that?
| Reporter | ||
Comment 15•5 years ago
•
|
||
Hi Dão,
I'm currently busy with writing up my masters' thesis (related to conic gradients in Firefox), would you mind taking care of potential necessary follow ups for this bug ?
Cheers,
Tim
Comment 16•5 years ago
|
||
Gonna have to back this out to address the regressions. Patch in bug 1634865.
Updated•5 years ago
|
| Reporter | ||
Updated•5 years ago
|
Updated•5 years ago
|
| Reporter | ||
Updated•5 years ago
|
Updated•3 years ago
|
Description
•