Fwd: Unittests: kcalendartest.cpp

[Jon Severinsson]

At Thursday Monday 26 November 2012, 10:47:48 David Faure wrote:
> One failure with Qt4:
> 
> FAIL!  : KCalendarTest::testHebrewStrings() Compared values are not the same
>    Actual (calendar->readDate(calendar->formatDate(testDate, KLocale::LongDate), KLocale::NormalFormat)): Invalid QDate
>    Expected (testDate): 2004/09/16
>    Loc: [/d/kde/src/5/kdelibs-frameworks/kdecore/tests/kcalendartest.cpp(804)]
> 
> John or Jon, could you have a look? :-)

At Thursday 13 December 2012 14:00:45 David Faure wrote:
> This test is still failing, with both Qt4 and Qt5 (more failures with Qt5
> though). Could you take a look at it?

The problem is that round tripping readDate() and formatDate() no longer work
when Hebrew numbers are involved.  I have isolated the cause down to the ki18n
split. If I revert the split [1] then this test passes on Qt4. I have yet to
isolate the actuall change that caused this, but I will look into it further
as I get some spare time if no one beats me to it (Chusslove?).

The other Qt5 failures are well known to me, and are fixed by my local patches
[2] (which I previously submitted to the list with little response). I have
started to push them through reviewboard (though my first four patches got some
heavy resistance by John Layt, more input would be welcome).

I have not prioritized getting them fixed up and merged yet, mostly because the
last patch, that actually fixes Qt5, also breaks Qt4. However, if we are about
to drop Qt4 support I guess the time to get it done is now :)

[1]: git://github.com/jonseverinsson/kdelibs.git revert-ki18n-split
[2]: git://github.com/jonseverinsson/kdelibs.git frameworks

> I'd like for all tests to pass with Qt5 before dropping Qt4 (and yes,
> waiting for 5.0.0 will probably be part of the equation too, although I
> suspect we'll need stuff from the "dev" branch pretty soon anyway)

Passing all tests are still way off, at least in my test environment (an LXC
container with no trace of KDE4). Currently the following tests fail on Qt5:
          4 - kdecore-kmountpointtest (Failed)
          8 - kdecore-kshelltest (Failed)
         11 - kdecore-kdatetimetest (Failed)
         18 - kdecore-ktcpsockettest (Failed)
         22 - kdecore-kcalendartest (Failed)
         82 - kio-krununittest (Failed)
         84 - kio-jobtest (Failed)
         86 - kio-jobremotetest (Failed)
         91 - kio-kdirmodeltest (Failed)
         92 - kio-kdirlistertest (Failed)
         93 - kio-fileundomanagertest (Failed)
         99 - kmimeassociationstest (Failed)
        100 - kfile-kfileplacesmodeltest (Failed)
        104 - kcookiejartest (Failed)
        118 - plasma-plasmoidpackagetest (Failed)
        119 - plasma-runnercontexttest (Failed)
        120 - plasma-storage (Failed)

But only the following tests fail on Qt4:
          5 - kdecore-kmountpointtest (Failed)
          9 - kdecore-kshelltest (Failed)
         12 - kdecore-kdatetimetest (Failed)
         23 - kdecore-kcalendartest (Failed)
         83 - kio-krununittest (Failed)
         85 - kio-jobtest (Failed)
         87 - kio-jobremotetest (Failed)
         92 - kio-kdirmodeltest (Failed)
         93 - kio-kdirlistertest (Failed)
         94 - kio-fileundomanagertest (Failed)
        100 - kmimeassociationstest (Failed)
        107 - kcookiejartest (Failed)

I'll look into these further, some might be because I've not yet gotten around
to building and installing kde-runtime on my test system.

Regards
Jon Severinsson
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://mail.kde.org/pipermail/kde-frameworks-devel/attachments/20121214/811940f8/attachment.sig>