Bug 85596 – [Qt]ietestcenter/css3/flexbox tests fail on 32 bit

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

[Csaba Osztrogonác]

Some tests introduced in http://trac.webkit.org/changeset/116018 
fail on 32 bit with minor pixel differences:

ietestcenter/css3/flexbox/flexbox-align-end-001.htm
ietestcenter/css3/flexbox/flexbox-align-start-001.htm
ietestcenter/css3/flexbox/flexbox-flex-001.htm
ietestcenter/css3/flexbox/flexbox-flex-002.htm
ietestcenter/css3/flexbox/flexbox-flex-003.htm
ietestcenter/css3/flexbox/flexbox-groups-001.htm
ietestcenter/css3/flexbox/flexbox-groups-002.htm
ietestcenter/css3/flexbox/flexbox-groups-004.htm
ietestcenter/css3/flexbox/flexbox-layout-003.htm

[Csaba Osztrogonác]

I skipped them - http://trac.webkit.org/changeset/116082

[Dave Tharp]

*** Bug 85666 has been marked as a duplicate of this bug. ***

[Dave Tharp]

Created attachment 140601 [details]
Patch

[Tony Chang]

Comment on attachment 140601 [details]
Patch

Did you file a bug to change the tests upstream?  I thought we normally don't change upstream tests.

[Dave Tharp]

(In reply to comment #4)
> (From update of attachment 140601 [details])
> Did you file a bug to change the tests upstream?  I thought we normally don't change upstream tests.

The upstream tests are not changed.  I wrote (and updated for this bug) the reference tests.

[Csaba Osztrogonác]

Comment on attachment 140601 [details]
Patch

I tested the patch on Qt 32 bit (where these tests failed) and 4 tests still fail:
ietestcenter/css3/flexbox/flexbox-align-end-001.htm = IMAGE
ietestcenter/css3/flexbox/flexbox-align-start-001.htm = IMAGE
ietestcenter/css3/flexbox/flexbox-flex-002.htm = IMAGE
ietestcenter/css3/flexbox/flexbox-layout-003.htm = IMAGE

[Csaba Osztrogonác]

Created attachment 140730 [details]
Qt failing results

[Csaba Osztrogonác]

(In reply to comment #7)
> Created an attachment (id=140730) [details]
> Qt failing results

And all tests pass on Qt 64 bit (with and without this patch too)

[Dave Tharp]

(In reply to comment #8)
> (In reply to comment #7)
> > Created an attachment (id=140730) [details] [details]
> > Qt failing results
> 
> And all tests pass on Qt 64 bit (with and without this patch too)

Ossy, Thanks for testing this.  I'm on 64 bit, so that's why I'm not seeing a failure.  I'll see about getting a 32bit system up and running to repro.

[Dave Tharp]

Created attachment 140974 [details]
Patch

[Dave Tharp]

(In reply to comment #9)
> (In reply to comment #8)
> > (In reply to comment #7)
> > > Created an attachment (id=140730) [details] [details] [details]
> > > Qt failing results
> > 
> > And all tests pass on Qt 64 bit (with and without this patch too)
> 
> Ossy, Thanks for testing this.  I'm on 64 bit, so that's why I'm not seeing a failure.  I'll see about getting a 32bit system up and running to repro.

I was able to build a 32-bit test system and repro on QT32 bit.  I've reworked the ref tests to workaround the QT 32-bit strangeness, and verified everything is working properly on QT32 as well as Mac Lion (these were the two problem children, but as it turns out, for separate reasons).

I have created a bug which captures the Qt32 strangeness: bug 86026

[Csaba Osztrogonác]

Comment on attachment 140974 [details]
Patch

I tested this patch, it works fine on 32 and on 64 bit Qt, so r=me.

Please update qt/test-expectations.txt and mac/Skipped file
before landing, because they coflict with the trunk now.

[Tom Zakrajsek]

Committed r116784: <http://trac.webkit.org/changeset/116784>