best unittest result ever:

[Jos van den Oever]

On Wednesday, June 22, 2011 23:42:02 PM Marijn Kruisselbrink wrote:
> On Wednesday, June 22, 2011 02:32:21 pm Jos van den Oever wrote:
> > On Wednesday, June 22, 2011 23:25:55 PM Jos van den Oever wrote:
> > > On Wednesday, June 22, 2011 23:13:40 PM Jos van den Oever wrote:
> > > > On Wednesday, June 22, 2011 14:51:52 PM Cyrille Berger Skott wrote:
> > > > > On Wednesday 22 June 2011, Boudewijn Rempt wrote:
> > > > > > On Wednesday 22 June 2011 Jun, Cyrille Berger Skott wrote:
> > > > > > > > while on Jos' nameless build
> > > > > > > > server, many krita/pigment tests fail for some weird reason
> > > > > > > > (they shouldn't be enabled at all there, actually).
> > > > > > > 
> > > > > > > I guess some dependencies check are not working correctly. What
> > > > > > > is the missing library for that build server ?
> > > > > > 
> > > > > > I'm not sure, but it looks a bit like there isn't any lcms -- and
> > > > > > if lcms is absent, krita shouldn't be built:
> > > > > > 
> > > > > > ### Krita: pigment, eigen2, exiv2
> > > > > > if(NOT EIGEN2_FOUND OR NOT EXIV2_FOUND OR NOT
> > > > > > HAVE_REQUIRED_LCMS_VERSION)
> > > > > > 
> > > > > >    set(SHOULD_BUILD_KRITA FALSE)
> > > > > > 
> > > > > > endif(NOT EIGEN2_FOUND OR NOT EXIV2_FOUND OR NOT
> > > > > > HAVE_REQUIRED_LCMS_VERSION)
> > > > > 
> > > > > I actually think that Jos need to do some cleaning on his build
> > > > > server.
> > > > > 
> > > > > Because if you go to the report page and look at log, and show all
> > > > > message, you get this:
> > > > > 
> > > > > [13:05:50]: -- ---------------- The following applications will be
> > > > > built ----------------
> > > > > [13:05:50]: -- words
> > > > > [13:05:50]: -- tables
> > > > > [13:05:50]: -- kpresenter
> > > > > [13:05:50]: -- kformula
> > > > > [13:05:50]: -- mobile
> > > > > [13:05:50]: -- calligra
> > > > > [13:05:50]: -- cstester
> > > > > [13:05:50]: --
> > > > 
> > > > Whenever a build fails, calligra is checked out again and built from
> > > > scratch, so this log is intrueguing. I'll have a look at what might
> > > > be lying around.
> > > > 
> > > > Cheers,
> > > > Jos
> > > > _______________________________________________
> > > > calligra-devel mailing list
> > > > calligra-devel at kde.org
> > > > https://mail.kde.org/mailman/listinfo/calligra-devel
> > > 
> > > The build machine has /usr/lib/liblcms.so.1 but that's it. No other
> > > strange things that might trigger a krita build or test run as far as i
> > > can tell.
> > > 
> > > Cheers,
> > > Jos
> > > _______________________________________________
> > > calligra-devel mailing list
> > > calligra-devel at kde.org
> > > https://mail.kde.org/mailman/listinfo/calligra-devel
> > 
> > Here's something that might be interesting. This is how the tests are run
> > on the build server:
> > ===
> > #! /bin/bash
> > SCRIPTDIR=`dirname "$0"`
> > 
> > # free memory from remaining jobs
> > killall -u $USER kdeinit4; killall -u $USER knotify4; killall -u $USER
> > klauncher
> > 
> > for f in `find -name '*.shell'|grep -i test`; do
> > 
> >     TML=${f%.shell}.tml
> >     XML=${f%.shell}.xml
> >     rm -f $XML $TML
> >     timeout 30 $f -xml -o $TML
> >     xsltproc -o $XML ${SCRIPTDIR}/qtest2junit.xsl $TML
> > 
> > done
> > ===
> > So if the .shell files for krita tests are created, they are run.
> 
> Wouldn't parsing the output of ctest -V -N be a much more reliable method
> to figure out what tests need to be run?

for f in `ctest -V -N -O -|cut -f 4 -d ' '|grep $PWD`; do

is indeed a more exact way of getting the list of tests, implemented now.

Failed tests are down to 11.


-- 
Jos van den Oever, software architect
+49 391 25 19 15 53
074 3491911
http://kogmbh.com/legal/