1097437 - Corners of Quartz StrokeRects are slightly off color

Status: RESOLVED FIXED

(Core :: Graphics, defect)

Description

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Drawing a simple <div style="border:1px solid black"> on a white background on Mac gives us corner pixels that are rgb(1,1,1) instead of rgb(0,0,0).

Comment 1

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: Work around Quartz bug where corners of stroked rects don't get a solid color when they should

Comment 2

[Robert O'Callahan (:roc) (email my personal email if necessary)]

try showed a few issues:
http://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=http://ftp.mozilla.org/pub/mozilla.org/firefox/try-builds/[e-mail]-00e6553ea192/try-macosx64-debug/try_snowleopard-debug_test-reftest-bm106-tests1-macosx-build3407.txt.gz&only_show_unexpected=1

* a couple of UNEXPECTED-PASSes ... so this was in fact already biting us.
* we shouldn't disable antialiasing when the pattern is not a solid color (since that will affect resampling of gradients/surfaces)
* we should generalize the workaround slightly so that it handles rotations by multiples of 90 degrees

Comment 3

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: patch v2

https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=65dc50d0d21e

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: fix v3

https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=c902e584931f

Comment 5

[Robert O'Callahan (:roc) (email my personal email if necessary)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/2999f246a4e2

Comment 6

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/2999f246a4e2

Comment 7

[Lawrence Mandel [:lmandel] (use needinfo)]

Comment on attachment 8521198 [details] [diff] [review]
fix v3

We need to uplift this patch to Beta and Aurora in support of bug 1092842. Roc tells me that this is low risk. Certainly seems worth taking two low risk fixes as opposed to shipping a visible YouTube regression. Beta+ Aurora+

Comment 8

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/88c5eac1e0e5
https://hg.mozilla.org/releases/mozilla-beta/rev/9e4c3c78fe01

Comment 9

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Looks like those unexpected passes from comment 2 just showed up on aurora and beta following that landing.

Comment 10

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I just landed the **reftest manifest changes only** from bug 1096181 on aurora and beta, because those changes actually belonged in either this bug or bug 1092842, which were just landed in both places:

https://hg.mozilla.org/releases/mozilla-aurora/rev/187e9fa68b67
https://hg.mozilla.org/releases/mozilla-beta/rev/c486cd17bebb