Obscure border-radius bug on Macs when one video plays on top of another
Categories
(Core :: Graphics: WebRender, defect)
Tracking
()
People
(Reporter: bugzilla, Unassigned)
Details
Attachments
(1 file)
|
3.78 KB,
text/html
|
Details |
User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:120.0) Gecko/20100101 Firefox/120.0
Steps to reproduce:
Produced a web page where video 1, which has a border-radius to round its corners, is playing in front of video 2.
Actual results:
In Firefox on the Mac (both my old Intel Mac and my newer M1 Mac), the corners of video 1 spill across the rounded border. This does not happen in Safari or Chrome; nor does it happen in Firefox running on Windows. The problem also goes away if video 2 is removed from behind video 1, or if a filter is applied to video 1 such as "hue-rotate(0.001deg)"...perhaps there's some sort of hardware vs. software video-rendering heuristic that needs adjustment?
Expected results:
Video 1 should have stayed within the rounded border.
|
||
Comment 1•1 year ago
|
||
The Bugbug bot thinks this bug should belong to the 'Core::Audio/Video: Playback' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.
Actually, it looks like I can see the buggy behavior sometimes in Firefox 121.0 (64-bit) on Windows -- that is, it's not a problem exclusively on the Mac.
The attached example HTML page works just fine (no bug visible) in my Windows Firefox if it's loaded in its own tab, but if it's displayed in an iframe embedded in a chat website I use, then the bug shows up. Yikes, that's super-weird.
|
||
Comment 3•1 year ago
|
||
I cant repro by just opening the attached HTML. The bug sounds vaguely similar to bug 1869994 (and other regressions from bug 1849680). If you could create a standalone testcase that is easier to reproduce, it would be quite helpful.
Can you perform a bisection using the mozregression tool (https://mozilla.github.io/mozregression/ )
|
|
||
Updated•1 year ago
|
|
||
Comment 4•1 year ago
|
||
Thanks for the reduced testcase.
I get this fix range
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=196748365c175513e46aa25757f5916566273639&tochange=e96bd5ec21a297e0521819a52b24aa05f44ce66d
-> bug 1869994
It should be fixed in the next version of Firefox (122).
I'm going to mark this duplicate of that bug, feel free to post if you have any other questions.
Description
•