Closed Bug 1316783 Opened 3 years ago Closed 3 years ago
[e10s] box-shadow is incorrectly tiled when printed
312.00 KB, application/pdf
4.48 KB, patch
|Details | Diff | Splinter Review|
I can reproduce this on OS X and Windows with the latest Nightly (As well as one from about a week ago), trying to print the linked page shows the shadows in the "Live examples" section being tiled instead of displayed normally. Both text-shadow and drop-shadow() display fine, the issue's only with box-shadow.
[Tracking Requested - why for this release]: Print output is broken against box-shadow Reproduce on Windows10 Nightly52.0a1, Aurora51.0a2 and 50.0rc2. Not reproduce on Windows10 49.0.2.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Version: Trunk → 50 Branch
Disable e10s seems to fix the box-shadow problem.
Summary: box-shadow is incorrectly tiled when printed → [e10s] box-shadow is incorrectly tiled when printed
Regression window: https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=fa802cc2231d1933e64c2e27fcbcd17394563891&tochange=69e6782758af0cccc4cfd4bae482e4d5c61ff0da Regressed by: 69e6782758af Bob Owen — Bug 1156742 Part 25: Flip the big switch and wait for the lightning. r=mconley
Track 51+/52+ as regression related to print output.
Confirmed the issue is related to e10s and can be avoided when setting 'print.print_via_parent' to false. After talking to GFX team Ethan Lin, there might be a problem when recording draw commands for box-shadow to printing DrawTarget.
I think I found the problem. Looks like it doesn't record the mSamplingRect for the pattern used to extend the shadow.
Assignee: nobody → bobowencode
Status: NEW → ASSIGNED
Attachment #8810386 - Flags: review?(mstange)
Attachment #8810347 - Attachment is obsolete: true
Attachment #8810386 - Flags: review?(mstange) → review+
We can uplift this fix to 50.1.0 if deemed safe. Andrew Overholt and I have reviewed this bug. It isn't release blocking for 50.0 go live.
Pushed by email@example.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/0e267219153d Record mSamplingRect for PatternType::SURFACE when using DrawTargetRecording. r=mstange
Please request Aurora/Beta approval on this when you get a chance. And Release if you think it's worth trying to get into December's 50.1 release.
Comment on attachment 8810386 [details] [diff] [review] Record mSamplingRect for PatternType::SURFACE when using DrawTargetRecording Approval Request Comment [Feature/regressing bug #]: Bug 1156742 (which uncovered an existing issue in the DrawTargetRecording). [User impact if declined]: box-shadow will continue to print incorrectly. [Describe test coverage new/current, TreeHerder]: Manual testing only. Tested with MDN page from bug report and two example pages linked from that page: https://developer.mozilla.org/en-US/docs/Web/CSS/box-shadow http://www.elektronotdienst-nuernberg.de/bugs/box-shadow_inset.html http://markusstange.wordpress.com/2009/02/15/fun-with-box-shadows/ [Risks and why]: low - fairly simple change to record the extra parameter required. The only concern is the lack of automated testing. [String/UUID change made/needed]: None
Comment on attachment 8810386 [details] [diff] [review] Record mSamplingRect for PatternType::SURFACE when using DrawTargetRecording Fix an issue related to printing. Beta51+ and Aurora52+. Should be in 51 beta 2.
Hello Alex, could you please verify this issue is fixed as expected on a latest Nightly build? Thanks!
Oh sorry, it's working fine now on everything I've tested with.
(In reply to Alex from comment #18) > Oh sorry, it's working fine now on everything I've tested with. Thanks!
Status: RESOLVED → VERIFIED
Comment on attachment 8810386 [details] [diff] [review] Record mSamplingRect for PatternType::SURFACE when using DrawTargetRecording Recent regression in 50, this wasn't deemed release blocking but it seems to be a good bug to fix in 50.1.0
Attachment #8810386 - Flags: approval-mozilla-release? → approval-mozilla-release+
(In reply to Alex from comment #18) > Oh sorry, it's working fine now on everything I've tested with. It seems that no additional qa work is needed here. Removing flag.
You need to log in before you can comment on or make changes to this bug.