Bug 92251 - Buildmaster changes in preparation of a Chromium Android tester
Summary: Buildmaster changes in preparation of a Chromium Android tester
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Peter Beverloo
URL:
Keywords:
Depends on:
Blocks: 84843
  Show dependency treegraph
 
Reported: 2012-07-25 07:24 PDT by Peter Beverloo
Modified: 2012-08-08 12:11 PDT (History)
4 users (show)

See Also:


Attachments
Patch (6.86 KB, patch)
2012-07-25 07:31 PDT, Peter Beverloo
no flags Details | Formatted Diff | Diff
Patch (7.85 KB, patch)
2012-07-25 08:25 PDT, Peter Beverloo
no flags Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Peter Beverloo 2012-07-25 07:24:57 PDT
Buildmaster changes in preparation of a Chromium Android tester
Comment 1 Peter Beverloo 2012-07-25 07:31:49 PDT
Created attachment 154350 [details]
Patch
Comment 2 Peter Beverloo 2012-07-25 07:35:58 PDT
The RunChromiumWebKitUnitTests, RunWebKitTests and RunUnitTests (which is TestWebKitAPI) steps are still broken, but the modifications required there can be made on the scripts' side and won't require master modifications.

After this and fixing the steps, all that remains should be adding the bot itself. Hardware is standing by, devices should be available shortly.

Lucas: ETA for adding the "Chromium Android Release (Tests)" bot should be roughly one to two weeks, would this work for you? Longer-term, say 1-2 months, we'd also like to add a "Chromium Android Release (Perf)" bot.
Comment 3 Adam Barth 2012-07-25 08:22:00 PDT
Comment on attachment 154350 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=154350&action=review

> Tools/BuildSlaveSupport/built-product-archive:172
> +        ignorePatterns = ['.svn', '*.a', '*.d', '*.dSYM', '*.o', '*.ilk', '*.lib', '*.idb',
> +                          'BuildLog.htm', '*.obj', '*.pdb', '*.pch', '*.tlog', '*.lastbuildstate']
> +
> +        if fullPlatform and fullPlatform == 'chromium-android':
> +            ignorePatterns.extend(['*.so', '*.pak', '*-unaligned.apk'])

Should we just ignore these files all the time?  For example, we ignore both *.obj and *.o even though those never exist on the same OS.
Comment 4 Peter Beverloo 2012-07-25 08:25:03 PDT
Created attachment 154361 [details]
Patch
Comment 5 Peter Beverloo 2012-07-25 08:27:53 PDT
(In reply to comment #3)
> Should we just ignore these files all the time?  For example, we ignore both *.obj and *.o even though those never exist on the same OS.

I answered that in the ChangeLog (maybe too verbose..?): at least the Chromium Mac and Linux builders contain .so files in their out directory, i.e. for ffmpegsudo. Since it's there, I've *assumed* that the libraries are being used.

*.pak and *-unaligned.apk would be safe to add to the general list, but since we have to make this separation anyway I thought it would be clearer to add them here.
Comment 6 Adam Barth 2012-07-25 09:01:25 PDT
Comment on attachment 154361 [details]
Patch

I fail at reading.  Thanks.
Comment 7 Peter Beverloo 2012-08-08 11:15:04 PDT
Lucas, do you have comments about this? Otherwise I'll put it on the CQ.
Comment 8 Lucas Forschler 2012-08-08 11:19:49 PDT
Comment on attachment 154361 [details]
Patch

I think this should be ok, fire away.
Comment 9 Peter Beverloo 2012-08-08 11:22:19 PDT
Comment on attachment 154361 [details]
Patch

Ok, thank you!
Comment 10 WebKit Review Bot 2012-08-08 12:11:52 PDT
Comment on attachment 154361 [details]
Patch

Clearing flags on attachment: 154361

Committed r125066: <http://trac.webkit.org/changeset/125066>
Comment 11 WebKit Review Bot 2012-08-08 12:11:56 PDT
All reviewed patches have been landed.  Closing bug.