Bug 43543 - Chromium: isindex tests timeout after r64712
Summary: Chromium: isindex tests timeout after r64712
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebKit Misc. (show other bugs)
Version: 528+ (Nightly build)
Hardware: All All
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-08-05 01:43 PDT by Pavel Feldman
Modified: 2010-08-08 00:33 PDT (History)
4 users (show)

See Also:


Attachments
Faster failing isindex tests (4.64 KB, patch)
2010-08-07 11:33 PDT, Mihai Parparita
no flags Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Pavel Feldman 2010-08-05 01:43:00 PDT
http/tests/misc/isindex-formdata.html = TIMEOUT
http/tests/misc/isindex-with-no-form-base-href.html = TEXT TIMEOUT
Comment 1 Pavel Feldman 2010-08-05 08:17:31 PDT
More timeout suppressions:

BUGWK43543 DEBUG : fast/parser/residual-style-dom.html = TIMEOUT
BUGWK43543 LINUX MAC DEBUG : fast/parser/residual-style-hang.html = TIMEOUT

Committing to http://svn.webkit.org/repository/webkit/trunk ...
	M	LayoutTests/ChangeLog
	M	LayoutTests/platform/chromium/test_expectations.txt
Committed r64751
Comment 2 Mihai Parparita 2010-08-07 11:33:45 PDT
Created attachment 63821 [details]
Faster failing isindex tests
Comment 3 Adam Barth 2010-08-07 11:38:20 PDT
Comment on attachment 63821 [details]
Faster failing isindex tests

Thanks.
Comment 4 Mihai Parparita 2010-08-07 11:41:27 PDT
I happened to be running all the layout tests with --slowest, and I noticed that isindex-with-no-form-base-href.html and isindex-formdata.html both took 30 seconds, since they trigger a DRT timeout with the HTML5 builder. The attached patch still has them fail, but in <1 second (since it actually checks for the things that are no longer there with the HTML 5 parser). Is something like this worth committing as a stopgap?

As for the tests themselves, isindex-formdata.html is definitely no longer valid with the HTML5 tree builder, since <isindex> elements inside <form> is supposed to be ignored (and is tested by LayoutTests/html5lib/resources/isindex.dat), so perhaps it could be deleted altogether.

isindex-with-no-form-base-href.html seems tricker, since it has a naked <isindex> element, which per http://www.whatwg.org/specs/web-apps/current-work/multipage/tokenization.html#isindex should generate a wrapping <form> (and does). However, that form has no action attribute, and I guess pre-HTML5 builder, the generated (empty string) form action was resolved against the <base href>, but that's no longer the case. I can't find anything in the HTML5 spec that defines the behavior either way, but Adam or Eric are probably more familiar with that.
Comment 5 Adam Barth 2010-08-07 11:51:32 PDT
I think it's worth landing.  The results of these tests match Minefield and the spec, as far as I can tell.
Comment 6 Eric Seidel (no email) 2010-08-08 00:33:45 PDT
Comment on attachment 63821 [details]
Faster failing isindex tests

Clearing flags on attachment: 63821

Committed r64944: <http://trac.webkit.org/changeset/64944>
Comment 7 Eric Seidel (no email) 2010-08-08 00:33:51 PDT
All reviewed patches have been landed.  Closing bug.