[Libosinfo] [libosinfo PATCH] tests: convert from check to the GLib testing framework
Guido Günther
agx at sigxcpu.org
Fri Oct 20 07:16:21 UTC 2017
Hi,
On Thu, Oct 19, 2017 at 04:14:14PM +0200, Pino Toscano wrote:
> On Thursday, 19 October 2017 16:02:18 CEST Daniel P. Berrange wrote:
> > On Thu, Oct 19, 2017 at 03:59:02PM +0200, Pino Toscano wrote:
> > > On Thursday, 19 October 2017 15:36:28 CEST Daniel P. Berrange wrote:
> > > > On Thu, Oct 19, 2017 at 03:14:24PM +0200, Pino Toscano wrote:
> > > > > diff --git a/tests/test-isodetect.c b/tests/test-isodetect.c
> > > > > index 7214531..97c86c4 100644
> > > > > --- a/tests/test-isodetect.c
> > > > > +++ b/tests/test-isodetect.c
> > > >
> > > > > -int main(void)
> > > > > -{
> > > > > - int number_failed;
> > > > > - Suite *s = list_suite();
> > > > > - SRunner *sr = srunner_create(s);
> > > > > + g_test_init(&argc, &argv, NULL);
> > > > > +
> > > > > + g_test_add_func("/isodetect/fedora", test_fedora);
> > > > > + g_test_add_func("/isodetect/rhel", test_rhel);
> > > > > + g_test_add_func("/isodetect/ubuntu", test_ubuntu);
> > > > > + g_test_add_func("/isodetect/debian", test_debian);
> > > > > + g_test_add_func("/isodetect/windows", test_windows);
> > > > > + g_test_add_func("/isodetect/freebsd", test_freebsd);
> > > > > + g_test_add_func("/isodetect/openbsd", test_openbsd);
> > > > > + g_test_add_func("/isodetect/opensuse", test_opensuse);
> > > > > + g_test_add_func("/isodetect/centos", test_centos);
> > > > > + g_test_add_func("/isodetect/gnome", test_gnome);
> > > > > + g_test_add_func("/isodetect/altlinux", test_altlinux);
> > > > > + g_test_add_func("/isodetect/mageia", test_mageia);
> > > > > + g_test_add_func("/isodetect/sles", test_sles);
> > > > > + g_test_add_func("/isodetect/sled", test_sled);
> > > >
> > > > We could wrap all of this in a
> > > >
> > > > if (g_test_slow()) {
> > > > ....
> > > > }
> > > >
> > > > so that this only executes if people request slow tests via '-m slow' arg
> > >
> > > Hm aren't these tests worth executing by default? Not sure what is the
> > > proper definition of "slow" for GLib tests though, and test-isodetect
> > > takes less than 20 seconds to run here.
> >
> > They're primarily useful for upstream to run - I don't think there's much
> > benefit downstream users running them. So I think its the kind of thing
> > where we could put '-m slow' into our CI system tests.
>
> Couldn't it be useful for downstream as well, i.e. to check that the
> data provided in osinfo-db actually match the expected ISO data? Say
> in case a downstream adds custom changes (or backport upstream commits),
> so they know the modified osinfo-db is OK.
>
> > THis would avoid downstream users needing networking during tests too IIUC.
>
> Note that test-isodetect does not require networking, as it uses the
> data in tests/isodata.
>From a distro point of view I'd rather have these run by default and see
-mslow preserved for really slow tests.
Cheers,
-- Guido
More information about the Libosinfo
mailing list