Re: Build/test passes locally but fails on hydra?

[Mark H Weaver]

Hi Hartmut,

Hartmut Goebel <address@hidden> writes:

> I wonder how it can happen that a build resp. test passes locally but
> fails on hydra. I'd assume on both machines the containers are set up
> the same. Any ideas?

One notable difference is the kernel.  Hydra.gnunet.org, the build slave
that performed the failed build below, and currently the only functional
Intel build slave on Hydra, is running a 3.16.x kernel.  Perhaps QtTest
is depending on functionality that's only available in newer kernels?

> Specific it is kdelibs4support-5.49.0 address@hidden 05:28:26}
> <https://hydra.gnu.org/build/3296401> where the test shown below fails
> at hydra. The test passes on my local machine.
>
> 33/37 Test #33: kglobalsettingstest ......................***Failed   10.14 
> sec
> QStandardPaths: XDG_RUNTIME_DIR not set, defaulting to 
> '/tmp/guix-build-kdelibs4support-5.49.0.drv-0/runtime-nixbld'
> ********* Start testing of KGlobalSettingsTest *********
> Config: Using QtTest library 5.11.2, Qt 5.11.2 (x86_64-little_endian-lp64 
> shared (dynamic) release build; by GCC 5.5.0)
> PASS   : KGlobalSettingsTest::initTestCase()
> FAIL!  : KGlobalSettingsTest::testPaletteChange() 'spy.wait(5000)' returned 
> FALSE. ()

I would start by investigating what 'spy.wait' does, and how it works.

    Regards,
      Mark