Bug 45610 – post-basic.html and post-frames.html in http/tests/navigation should not be pixel tests

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Mihai Parparita]

post-basic.html and post-frames.html in http/tests/navigation should not be pixel tests

[Mihai Parparita]

Created attachment 67330 [details]
Patch

[Mihai Parparita]

Dimitri? Should be very similar to bug 45307. Let me know if I should convert all the tests in that directory to not be pixel tests in one go (though then the patch might be unwieldy).

[Dimitri Glazkov (Google)]

Comment on attachment 67330 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=67330&action=prettypatch

weee!

[WebKit Commit Bot]

Comment on attachment 67330 [details]
Patch

Rejecting patch 67330 from commit-queue.

Failed to run "[u'/Users/eseidel/Projects/CommitQueue/WebKitTools/Scripts/svn-apply', u'--reviewer', u'Dimitri Glazkov', u'--force']" exit_code: 255
Last 500 characters of output:
urces/postresult.pl
patching file LayoutTests/platform/chromium-linux/http/tests/navigation/post-basic-expected.checksum
rm 'LayoutTests/platform/chromium-linux/http/tests/navigation/post-basic-expected.checksum'
literal is only avaliable with the XS version at /Library/Perl/5.8.8/Compress/Zlib.pm line 9
BEGIN failed--compilation aborted at /Library/Perl/5.8.8/Compress/Zlib.pm line 9.
Compilation failed in require at /Users/eseidel/Projects/CommitQueue/WebKitTools/Scripts/VCSUtils.pm line 1670.

Full output: http://queues.webkit.org/results/3964423

[Dimitri Glazkov (Google)]

Comment on attachment 67330 [details]
Patch

Clearing flags on attachment: 67330

Committed r67328: <http://trac.webkit.org/changeset/67328>

[Dimitri Glazkov (Google)]

All reviewed patches have been landed.  Closing bug.

[James Robinson]

Reverted r67328 for reason:

Broke

Committed r67349: <http://trac.webkit.org/changeset/67349>

[James Robinson]

Looks like the render tree expectations were lacking one 0x0 RenderText.  My feeble attempt to rebaseline the platform/mac and chromium expectations was a complete scripting failure so I rolled this out.

[Eric Seidel (no email)]

I believe this caused persistant failures on Mac Leopard.

[Eric Seidel (no email)]

I see, this was rolled out, thanks James!

[Mihai Parparita]

James, thanks for the rollout. You said in #webkit "also it's really unclear to me why those tests have render tree dumps. that seems wrong" Agreed, that's what I've been trying to do, convert the tests in that directory to just use dumpAsText. I'm guessing the reason why post-goback1.html and other tests started to fail (even though I didn't touch them) is because I added a <script> tag to postresult.pl, which is used by those tests too. I'm guessing that showed up as the 0x0 render tree object in the output.

I think the safest thing to do in the redone version of this patch is to branch postresult.pl, so that I don't inadvertedly affect other tests in the directory. Eventually (by the end of the week, hopefully), I hope to have all the tests converted, and then I can get rid of the original postresult.pl (as well as a lot of other stuff in this directory).

Incidentally, the cq should've caught those other test failures, but this couldn't be landed by the cq because of patch apply failures (comment 4). Eric, is this because of svn-apply being unable to handle git binary diffs (even for deletes?), or something else?

[Mihai Parparita]

Created attachment 67416 [details]
Patch

New version of the patch that doesn't touch postresult.pl

[Dimitri Glazkov (Google)]

Comment on attachment 67416 [details]
Patch

Yes!

[WebKit Commit Bot]

Comment on attachment 67416 [details]
Patch

Rejecting patch 67416 from commit-queue.

Failed to run "[u'/Users/eseidel/Projects/CommitQueue/WebKitTools/Scripts/svn-apply', u'--reviewer', u'Dimitri Glazkov', u'--force']" exit_code: 255
Last 500 characters of output:
age-that-posts.html
patching file LayoutTests/platform/chromium-linux/http/tests/navigation/post-basic-expected.checksum
rm 'LayoutTests/platform/chromium-linux/http/tests/navigation/post-basic-expected.checksum'
literal is only avaliable with the XS version at /Library/Perl/5.8.8/Compress/Zlib.pm line 9
BEGIN failed--compilation aborted at /Library/Perl/5.8.8/Compress/Zlib.pm line 9.
Compilation failed in require at /Users/eseidel/Projects/CommitQueue/WebKitTools/Scripts/VCSUtils.pm line 1670.

Full output: http://queues.webkit.org/results/4024002

[Mihai Parparita]

I filed bug 45723 about the cq bot error.

Dimitri, mind landing this?

[Eric Seidel (no email)]

Comment on attachment 67416 [details]
Patch

I attempted to fix the CQ bot just now.  Hopefully I was successful.  bug 45723.

[WebKit Commit Bot]

Comment on attachment 67416 [details]
Patch

Rejecting patch 67416 from commit-queue.

Failed to run "[u'/Users/eseidel/Projects/CommitQueue/WebKitTools/Scripts/svn-apply', u'--reviewer', u'Dimitri Glazkov', u'--force']" exit_code: 255
Last 500 characters of output:
age-that-posts.html
patching file LayoutTests/platform/chromium-linux/http/tests/navigation/post-basic-expected.checksum
rm 'LayoutTests/platform/chromium-linux/http/tests/navigation/post-basic-expected.checksum'
literal is only avaliable with the XS version at /Library/Perl/5.8.8/Compress/Zlib.pm line 9
BEGIN failed--compilation aborted at /Library/Perl/5.8.8/Compress/Zlib.pm line 9.
Compilation failed in require at /Users/eseidel/Projects/CommitQueue/WebKitTools/Scripts/VCSUtils.pm line 1670.

Full output: http://queues.webkit.org/results/4060005

[Eric Seidel (no email)]

Comment on attachment 67416 [details]
Patch

I believe it's fixed for real this time.

[WebKit Commit Bot]

Comment on attachment 67416 [details]
Patch

Clearing flags on attachment: 67416

Committed r67466: <http://trac.webkit.org/changeset/67466>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[WebKit Review Bot]

http://trac.webkit.org/changeset/67466 might have broken GTK Linux 32-bit Debug
The following changes are on the blame list:
http://trac.webkit.org/changeset/67465
http://trac.webkit.org/changeset/67466
http://trac.webkit.org/changeset/67467
http://trac.webkit.org/changeset/67468

[Mihai Parparita]

(In reply to comment #21)
> http://trac.webkit.org/changeset/67466 might have broken GTK Linux 32-bit Debug
> The following changes are on the blame list:
> http://trac.webkit.org/changeset/67465
> http://trac.webkit.org/changeset/67466
> http://trac.webkit.org/changeset/67467
> http://trac.webkit.org/changeset/67468

The new failure is that fast/dom/HTMLScriptElement/script-decoding-error-after-setting-src.html crashes, which I doubt was caused by my test-only change. r67467 (to fix bug 45649) seems like a more probable culprit.