breaking unittests is BAD

[Boudewijn Rempt]

Hi,

_If_ we want to be able to release master as a reasonably stable snapshot, we really have to be better about our unittests. Even if the unittests test something that is being refactored anyway in a branch, we should _NOT_ break those tests.


Currently we have the following failures:

          7 - words-part-frame-TestDocumentLayout (Failed)

This one is a crash after a refactoring by Sebastian, and when I remove the cause for the crash, the test fails completely anyway.

         10 - words-part-TestRdf (Failed)
         32 - kspread-Formula (Failed)
         63 - krita-image-KisIteratorTest (Failed)
         88 - krita-image-KisFilterTest (Failed)
        168 - krita-libbrush-KisAutobrushTest (Failed)
        176 - krita-ui-kis_exr_test (Failed)
        183 - libs-widgets-zoomhandler_test (Failed)

Arjen: you were going to fix this test -- please do that today!

        237 - libs-pigment-TestKoCompositeOps (Failed)
        266 - shapes-text-TestDocumentLayout (Failed)
        267 - shapes-text-TestChangeListCommand (Failed)
        269 - shapes-text-TestTableLayout (Failed)

And these three basically mean that we don't test text layout anymore.

        273 - kspread-xlsx-FormulaParser (Failed)

The output on http://my.cdash.org/index.php?project=Calligra isn't as nice as the teamcity setup, but people complained about that as well. I'm not sure how I can figure out who committed the changes that broke the test -- Cyrille, do you know how I can figure that out?

-- 
Boudewijn Rempt | http://www.valdyas.org, http://www.krita.org