Bug 76835 – http://crrev.com/118682 changed gradient layout test results

• Component: Tools / Tests
• Status: RESOLVED FIXED

My http://crrev.com/118682 (a Skia deps roll) changed various gradient layout test results.  (We believe the new results are better than the old ones, but for now it shows up as a regression.)

I included chrome test_expectations changes in that CL to keep the bots green until we rebaseline these tests, but apparently some of the chromium.webkit bots ignore that file.

I am working on a CL to add those tests to webkit's main test_expectations file, and then once all the deps rolls have taken place back and forth I will rebaseline.

[anton muhin]

+ Julie who is on PST shift

Created attachment 123568 [details]
Patch

[Dimitri Glazkov (Google)]

Comment on attachment 123568 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=123568&action=review

> LayoutTests/platform/chromium/test_expectations.txt:4016
> +BUGCR110926 : fast/backgrounds/repeat/negative-offset-repeat-transformed.html = PASS IMAGE TEXT IMAGE+TEXT

I don't think it means what you think it means :) You should just use FAIL here. As it is now, you're specifying that the test is rrreeeeallly flaky.

(In reply to comment #3)
> 
> I don't think it means what you think it means :) You should just use FAIL here. As it is now, you're specifying that the test is rrreeeeallly flaky.

Actually, here's what it means: It means that I have an Andre-The-Giant amount of work to get done, and I'm trying to work around the fact that we have this crazy setup of multiple test_expectations files (in webkit and chrome) in which *sometimes* they are both observed, and at *other* times only one is observed, and depending on when deps rolls happen in both directions it might take more than a week to properly rebaseline the tests.

This change mirrors my previous changes to the chrome test_expectations file, as closely as possible, so that these expectations as observed by ALL bots.  If I make the lines different here than in the chrome test_expectations file, it will just make it harder for people to understand.

This change should work; Anton and Julie, I will leave it to you to do whatever you want with it.  You can probably land it as-is if you want, or you can massage it to meet whatever abstract standards people wish to apply.  As always, I am busting my butt trying to get improvements out into production as quickly as possible while minimizing disturbances for other developers; I'm done with this particular issue.

assigning over to jparent

[Adam Barth]

epoger, I'm sorry you're frustrated by this process.  We probably should just delete the downstream test_expectations.txt file.  It seems to be providing negative value at this point.

Fixed in http://trac.webkit.org/changeset/105633.