Bug 68270 – non-CG WebkitMac buildbots have massive failures on Mac OS 10.5

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

We have recently turned on WebkitMac (non-CG) buildbots within http://build.chromium.org/p/chromium.webkit/waterfall .

The ones running on 10.6 are doing reasonably decent (although they are red--still failing a handful of tests).  But the ones running on 10.5 are failing 500+ tests.

This is probably due to the fact that Cary Clark and I don't have a decent 10.5 machine handy, so all of our local testing has been on 10.6.  We probably need to create a bunch of new baseline text/image results in http://trac.webkit.org/browser/trunk/LayoutTests/platform/chromium-mac-leopard

My plan, for now, is to modify test_expectations to allow lots of failures on 10.5.  Then I can rebaseline without all the redness.

Cary has landed my initial test_expectations change as http://trac.webkit.org/changeset/95317/

*** Bug 71935 has been marked as a duplicate of this bug. ***

Created attachment 114337 [details]
Patch

The patch I attached yesterday is the final step to re-enabling layout test comparisons on Leopard.  Looking for a reviewer...

I have marked this "commit-queue -" in an attempt to cancel... we had to roll out use_skia=1 in Chromium (yet again, sigh)

Oops, canceled the commit of the wrong patch. Set back to "commit-queue ?".

[WebKit Review Bot]

Comment on attachment 114337 [details]
Patch

Clearing flags on attachment: 114337

Committed r100008: <http://trac.webkit.org/changeset/100008>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.

[Ojan Vafai]

Not sure what happened here. The commit log is very confusing. In either case, http://trac.webkit.org/changeset/100008 caused hundreds of failures on 10.5 and epoger is not online. I'm rolling back.

[Ojan Vafai]

Rollout: http://trac.webkit.org/changeset/100044

(In reply to comment #9)
> Not sure what happened here. The commit log is very confusing. In either case, http://trac.webkit.org/changeset/100008 caused hundreds of failures on 10.5 and epoger is not online. I'm rolling back.

(sigh) Thanks for doing the rollback, Ojan.  The problem is that I clicked "commit +" on this and then it didn't get committed by the commit-bot until 1.5 days later (right before I left the office on Friday afternoon!)

I'll refresh the CL and then get someone to land it for me.

Finally put in the final rebaselines (and removed the broad test_expectations failures) in http://trac.webkit.org/changeset/100917 .