91677 – REGRESSION (r122660-r122663): mathml/presentation/mo.xhtml, mathml/presentation/row.xhtml failing on Mountain Lion Production Tests

Bug 91677 - REGRESSION (r122660-r122663): mathml/presentation/mo.xhtml, mathml/presentation/row.xhtml failing on Mountain Lion Production Tests

Summary: REGRESSION (r122660-r122663): mathml/presentation/mo.xhtml, mathml/presentati...

Status:	RESOLVED FIXED

Alias:	None

Product:	WebKit
Classification:	Unclassified
Component:	Tools / Tests (show other bugs)
Version:	528+ (Nightly build)
Hardware:	Unspecified Unspecified

Importance:	P2 Normal
Assignee:	Jer Noble

URL:	http://build.webkit.org/results/Apple...
Keywords:	LayoutTestFailure, MakingBotsRed, Regression

Depends on:
Blocks:

Reported:	2012-07-18 15:08 PDT by Jer Noble
Modified:	2012-07-18 15:58 PDT (History)
CC List:	4 users (show)

See Also:

Attachments
mathml/ LayoutTest results (146.13 KB, application/zip) 2012-07-18 15:30 PDT, Jer Noble	no flags	Details
Test Case (448 bytes, application/xhtml+xml) 2012-07-18 15:38 PDT, Robert Hogan	no flags	Details
Patch (130.21 KB, patch) 2012-07-18 15:43 PDT, Jer Noble	no flags	Details \| Formatted Diff \| Diff
Patch (220.61 KB, patch) 2012-07-18 15:49 PDT, Jer Noble	eric: review+ eric: commit-queue+	Details \| Formatted Diff \| Diff
Show Obsolete (1) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.

Description Jer Noble 2012-07-18 15:08:31 PDT

mathml/presentation/mo.xhtml and mathml/presentation/row.xhtml started failing on Apple Lion Release WK1 (Tests) between r122661 and r122663 (inclusive).

http://trac.webkit.org/log/trunk?rev=122663&stop_rev=122661&limit=4

http://build.webkit.org/results/Apple%20Lion%20Release%20WK1%20(Tests)/r122660%20(1138)/results.html passed
http://build.webkit.org/results/Apple%20Lion%20Release%20WK1%20(Tests)/r122663%20(1139)/results.html failed

Comment 1 Robert Hogan 2012-07-18 15:24:35 PDT

The Mac bots don't run pixel tests - would it be possible to see the image diff of these failing tests? (Chromium linux doesn't fully support mathml as far as I can tell).

Comment 2 Jer Noble 2012-07-18 15:27:09 PDT

Sure, let me run one from both revisions.

Comment 3 Jer Noble 2012-07-18 15:30:50 PDT

Created attachment 153107 [details]
mathml/ LayoutTest results

Comment 4 Robert Hogan 2012-07-18 15:36:31 PDT

The new layout is closer to Firefox - i.e. the baseline of each green box is near the middle of the parentheses. I think the change in result here is due to the test depending on the old, wrong behaviour of vertical-align with a percentage value.

Comment 5 Robert Hogan 2012-07-18 15:38:08 PDT

Created attachment 153111 [details]
Test Case

This shows the new behaviour of vertical-align with a negative percentage. WebKit's current rendering of this matches Firefox and Opera, whereas its old rendering does not.

Comment 6 Jer Noble 2012-07-18 15:43:52 PDT

Created attachment 153113 [details]
Patch

Comment 7 Robert Hogan 2012-07-18 15:44:10 PDT

Hi Jer,

So, I would say it's safe to rebaseline. Don't know if you want anyone else to chime in on it.

I'll land the new results tomorrow if no-one objects.

Thanks,
Robert

Comment 8 Eric Seidel (no email) 2012-07-18 15:47:39 PDT

Comment on attachment 153113 [details]
Patch

rs=me.

Comment 9 Jer Noble 2012-07-18 15:49:36 PDT

Created attachment 153117 [details]
Patch

Comment 10 Eric Seidel (no email) 2012-07-18 15:53:15 PDT

Comment on attachment 153117 [details]
Patch

rs=me.

Comment 11 Jer Noble 2012-07-18 15:58:54 PDT

Committed r123039: <http://trac.webkit.org/changeset/123039>