[webkit-qt] Maintaining Qt 4.7 and 4.8 expected results in parallel

[Ademar Reis]

2011/9/7 Osztrogonac Csaba <oszi at inf.u-szeged.hu>:
> Hi QtWebKit developers,
>
> I ran into a serious problem what we should solve without any pain.
>
> Here is a bug report with an r+ -ed patch:
> https://bugs.webkit.org/show_bug.cgi?id=55036
> [Qt] FontCache::createFontPlatformData() is broken, a default font is
> returned even if the family does not match

May also be relevant: https://bugreports.qt.nokia.com/browse/QTBUG-21036
(this one I tried to push for inclusion in Qt-4.8 as it impacts
QtWebKit, but had no success)

>
> After this patch ~5000 layout test result will differ between Qt 4.7 and
> 4.8,
> which is absolutly unmaintainable. Additionally WebKit2 builds with Qt 5,
> but
> qt-wk2 platform falls back to qt platform and not qt-4.8/qt platforms.
>
> I don't know how important is this patch (AFAIK it doesn't block QtWebKit2.2
> release),
> but I think the best solution is if we wait until Qt 4.8 relase and then we
> drop
> LayoutTests/platform/qt-4.8 platform and declare that now
> LayoutTests/platform/qt
> means Qt 4.8 results and Qt 4.7 layout test results aren't supported
> anymore.
>
> If we want to land this patch now, we have to
> - fix run-webkit-tests to be able handle qt-wk2/qt-4.8/qt fallback order (It
> isn't trivial fix.)
> - add +5000 Qt 4.8 specific layout test result into
> LayoutTests/platform/qt-4.8
> - maintain LayoutTests/platform/qt and LayoutTests/platform/qt-4.8 in
> parallel ( It is shocking big task !!! )
>
> What do you think about it? What should we do now?

If the change to Qt-4.8 doesn't introduce any crashes (or timeouts),
then +1 for waiting until qt-4.8 is released and dropping support for
qt-4.7 on layout tests.

Ditto for QtWebKit-2.2(.0).

Cheers,
   - Ademar

-- 
Ademar de Souza Reis Jr. <ademar.reis at openbossa.org>
Nokia Institute of Technology