[webkit-qt] qt5 version updated for bot / development

simon.hausmann at nokia.com simon.hausmann at nokia.com
Fri May 4 10:44:05 PDT 2012


Hi,


A lot of these regressions are also due to behavioral changes in Qt, but they still end up as regressions because
they require us to change the way we use Qt.

Ossy is right, this is getting way to big and we're running risk of destabilizing our foundations. At the same time
new features are being implemented at a rapid pace (maybe it's catching up with other ports or implementing
WebCore interfaces, but they're still new features).

We are now in week 18. How about we target week 20 for an concentrated effort of getting our test coverage back up? (that gives
a week to finish your pending features ;)

Who's in? :)



Simon

________________________________________
From: webkit-qt-bounces at lists.webkit.org [webkit-qt-bounces at lists.webkit.org] on behalf of ext Osztrogonac Csaba [oszi at inf.u-szeged.hu]
Sent: Friday, May 04, 2012 18:50
To: webkit-qt at lists.webkit.org
Subject: Re: [webkit-qt] qt5 version updated for bot / development

Hi,

Osztrogonac Csaba írta:
> Hi All,
>
> I updated the Qt5 on the Qt5 WK1/WK2 buildbots and on the Qt5 WK1/WK2
> performance bots.
>
> Additional buildfixes landed in WebKit trunk:
> - http://trac.webkit.org/changeset/116100
> - http://trac.webkit.org/changeset/116101
> - http://trac.webkit.org/changeset/116102
>
> You can find the updated build script here:
> https://github.com/ossy-szeged/qt5-tools
>
> New "weekly" Qt5 hash is 88442f81edb4464cc88ac7c970333e7e59f1d200 (Fri
> May 4 13:16:52 2012 +1000)
>
> And please update your testfonts too, it is necessarry to run Qt5-WK1
> layout tests.
>
> Unfortunately there are layout test regressions. :-/
> I'm filing bugs right now and will skip the failing tests.

I finished with filing bug reports and skipping tests to paper over the
new regressions: (and to make buildbots be able to catch new regressions!)
  - [Qt] REGRESSION: 10 tests started to fail with newer Qt5
    https://bugs.webkit.org/show_bug.cgi?id=85629
  - [Qt] REGRESSION: 350 tests started to fail with newer Qt5-WK1
    https://bugs.webkit.org/show_bug.cgi?id=85631
  - [Qt] REGRESSION: http/tests/xmlhttprequest/basic-auth-nopassword.html started to crash with newer Qt5
    https://bugs.webkit.org/show_bug.cgi?id=85632

Nowadays I have a strange feeling, that regressions/crashes in QtWebKit don't matter at all.
I'd be happy if I'm wrong, but the following meta bugs history cuts a poor figure:
  - meta bug for fixing crashes/assertions/timeouts with very very long
    depends on list: https://bugs.webkit.org/show_bug.cgi?id=79666
  - meta bug for fixing "lighter" regressions, which cause only layout test failures
    with very very long depends on list: https://bugs.webkit.org/show_bug.cgi?id=79666

br,
Ossy
_______________________________________________
webkit-qt mailing list
webkit-qt at lists.webkit.org
http://lists.webkit.org/mailman/listinfo.cgi/webkit-qt


More information about the webkit-qt mailing list