Bug 69429 - new-run-webkit-tests fails on Lion
Summary: new-run-webkit-tests fails on Lion
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: Tools / Tests (show other bugs)
Version: 528+ (Nightly build)
Hardware: Mac OS X 10.7
: P2 Normal
Assignee: epoger
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-10-05 08:16 PDT by epoger
Modified: 2011-10-06 07:43 PDT (History)
6 users (show)

See Also:


Attachments
Patch (3.31 KB, patch)
2011-10-05 08:31 PDT, epoger
no flags Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description epoger 2011-10-05 08:16:47 PDT
Right now, if you try to run this on Lion:

python ./Tools/Scripts/new-run-webkit-tests --no-show-results --no-new-test-results --exit-after-n-failures 9 --verbose --results-directory $OUTDIR --release --chromium --platform chromium-mac

it fails:

Traceback (most recent call last):
  File "/Volumes/epoger/src/webkit/run-layout-tests-on-lion/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py", line 440, in <module>
    sys.exit(main())
  File "/Volumes/epoger/src/webkit/run-layout-tests-on-lion/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py", line 435, in main
    return run(port, options, args)
  File "/Volumes/epoger/src/webkit/run-layout-tests-on-lion/WebKit/Tools/Scripts/webkitpy/layout_tests/run_webkit_tests.py", line 84, in run
    manager.print_config()
  File "/Volumes/epoger/src/webkit/run-layout-tests-on-lion/WebKit/Tools/Scripts/webkitpy/layout_tests/controllers/manager.py", line 1098, in print_config
    fallback_path = [self._fs.split(x)[1] for x in self._port.baseline_search_path()]
  File "/Volumes/epoger/src/webkit/run-layout-tests-on-lion/WebKit/Tools/Scripts/webkitpy/layout_tests/port/chromium_mac.py", line 114, in baseline_search_path
    return map(self._webkit_baseline_path, fallback_paths[self._version])
KeyError: 'lion'
Comment 1 epoger 2011-10-05 08:31:43 PDT
Created attachment 109799 [details]
Patch
Comment 2 epoger 2011-10-05 08:36:47 PDT
James- can you please review the attached patch?
Comment 3 James Robinson 2011-10-05 11:18:07 PDT
Eric, Adam, or Dirk would be better reviewers for this.
Comment 4 Adam Barth 2011-10-05 12:11:10 PDT
Comment on attachment 109799 [details]
Patch

Ok.  Do we need to create chromium-mac-snowleopard?  My understanding is that we won't have Lion bots until 10.7.2 due to an OS bug.
Comment 5 epoger 2011-10-05 14:03:13 PDT
(In reply to comment #4)
> (From update of attachment 109799 [details])
> Ok.  Do we need to create chromium-mac-snowleopard?  My understanding is that we won't have Lion bots until 10.7.2 due to an OS bug.

Yes, supposedly the above problem is detailed at http://code.google.com/p/chromium/issues/detail?id=87290 , but I don't have access to that bug. (Can someone please CC me on it?)

Right now I am making absolutely sure that layout tests still pass on SnowLeopard with this change.  Assuming they do, I will commit this patch tomorrow morning.
Comment 6 Adam Barth 2011-10-05 14:11:22 PDT
> Yes, supposedly the above problem is detailed at http://code.google.com/p/chromium/issues/detail?id=87290 , but I don't have access to that bug. (Can someone please CC me on it?)

Done.

> Right now I am making absolutely sure that layout tests still pass on SnowLeopard with this change.  Assuming they do, I will commit this patch tomorrow morning.

Great.
Comment 7 Eric Seidel (no email) 2011-10-05 15:19:40 PDT
Comment on attachment 109799 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=109799&action=review

We have the ability to unittest these fallback paths.

> Tools/Scripts/webkitpy/layout_tests/port/chromium_mac.py:61
> +            'chromium-mac-snowleopard',
>              'chromium-mac',

Why don't we have a chromium-lion in between chromium-mac and chromium-mac-snowleopard?  My understanding is that "mac" is normally the future OS, as I would assume "chromium-mac" would be whatever future OS.  Donno.
Comment 8 Dirk Pranke 2011-10-05 15:38:05 PDT
(In reply to comment #7)
> (From update of attachment 109799 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=109799&action=review
> 
> We have the ability to unittest these fallback paths.
> 
> > Tools/Scripts/webkitpy/layout_tests/port/chromium_mac.py:61
> > +            'chromium-mac-snowleopard',
> >              'chromium-mac',
> 
> Why don't we have a chromium-lion in between chromium-mac and chromium-mac-snowleopard?  My understanding is that "mac" is normally the future OS, as I would assume "chromium-mac" would be whatever future OS.  Donno.

For Chromium, at least today, "chromium-mac" is defined as "the latest shipping mac" release, since we're outside of apple and can only use the shipping products. So, "chromium-mac" == "lion", and there's no way for us to be able to manage "future" baselines separate from "lion" baselines (or even known when we'd need them). Only the people inside Apple can know this.

(Theoretically, if we someday got access to prerelease O/S releases, this could change).
Comment 9 WebKit Review Bot 2011-10-06 07:43:55 PDT
Comment on attachment 109799 [details]
Patch

Clearing flags on attachment: 109799

Committed r96811: <http://trac.webkit.org/changeset/96811>
Comment 10 WebKit Review Bot 2011-10-06 07:43:59 PDT
All reviewed patches have been landed.  Closing bug.