Bug 113811 - Need japanese line layout performance test
Summary: Need japanese line layout performance test
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: Tools / Tests (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Glenn Adams
URL:
Keywords:
Depends on:
Blocks: 105692
  Show dependency tree / graph
 
Reported: 2013-04-02 11:40 PDT by Glenn Adams
Modified: 2013-04-02 16:29 PDT (History)
4 users (show)

See Also:

Attachments
Patch (26.68 KB, patch)
2013-04-02 11:48 PDT, Glenn Adams 		no flags Details | Formatted Diff | Diff
run-perf-tests results (27.51 KB, text/html)
2013-04-02 14:56 PDT, Glenn Adams 		no flags Details
Patch (27.29 KB, patch)
2013-04-02 15:33 PDT, Glenn Adams 		no flags Details | Formatted Diff | Diff
Show Obsolete (1) View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Glenn Adams 2013-04-02 11:40:29 PDT 
For use in testing performance regression related to upcoming Japanese related line break patches.
Comment 1 Glenn Adams 2013-04-02 11:48:48 PDT 
Created attachment 196200 [details]
Patch
Comment 2 Ryosuke Niwa 2013-04-02 12:58:37 PDT 
Comment on attachment 196200 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=196200&action=review

> PerformanceTests/ChangeLog:9
> +        Will be used by https://bugs.webkit.org/show_bug.cgi?id=105692 to test potential performance regression
> +        for Japanese (and other CJK) line break changes.

Is this test stable? Could you post sample results from this test?
e.g. you can simply upload WebKitBuild/Release/PerformanceTestsResults.html after repeatedly running this test for 3-5 times.

Also, is it really worth running this test on bot? The current cycle time of bots is roughly 2 hours so I'd rather not enable this test by default unless this test is going to be very effective in detecting real regressions.

> PerformanceTests/Layout/LineLayoutJapanese.html:4
> +    <meta charset="utf-8" />

We don't explicitly close self-closing tag in HTML5.
Comment 3 Glenn Adams 2013-04-02 14:56:59 PDT 
Created attachment 196232 [details]
run-perf-tests results

Test run of PerformanceTests/Layout/LineLayoutJapanese.html on MBP Retina 15" (16GB RAM, 512G SSD).
Comment 4 Glenn Adams 2013-04-02 15:33:58 PDT 
Created attachment 196240 [details]
Patch
Comment 5 Glenn Adams 2013-04-02 15:36:02 PDT 
(In reply to comment #2)
> (From update of attachment 196200 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=196200&action=review
> 
> > PerformanceTests/ChangeLog:9
> > +        Will be used by https://bugs.webkit.org/show_bug.cgi?id=105692 to test potential performance regression
> > +        for Japanese (and other CJK) line break changes.
> 
> Is this test stable? Could you post sample results from this test?
> e.g. you can simply upload WebKitBuild/Release/PerformanceTestsResults.html after repeatedly running this test for 3-5 times.

Yes. Posted.

> 
> Also, is it really worth running this test on bot? The current cycle time of bots is roughly 2 hours so I'd rather not enable this test by default unless this test is going to be very effective in detecting real regressions.

Added to Skipped.

> 
> > PerformanceTests/Layout/LineLayoutJapanese.html:4
> > +    <meta charset="utf-8" />
> 
> We don't explicitly close self-closing tag in HTML5.

Fixed.
Comment 6 WebKit Review Bot 2013-04-02 16:29:42 PDT 
Comment on attachment 196240 [details]
Patch

Clearing flags on attachment: 196240

Committed r147507: <http://trac.webkit.org/changeset/147507>
Comment 7 WebKit Review Bot 2013-04-02 16:29:45 PDT 
All reviewed patches have been landed.  Closing bug.