On Fri, Jul 1, 2011 at 3:57 PM, Patrick Ohly <patrick.ohly(a)intel.com> wrote:
On Fri, 2011-07-01 at 12:24 +0200, Chris Kühl wrote:
> Would it be possible for you to make a step by step list of what you
> do to set up your environment for testing? I think this might help us
> find why we are getting different results for the "defaultPeer" issue,
> for example.
The only setup required in the host is the configuration of the
"dbus_unittest" sync config. I added some comments about that to the
TestSessionAPIsReal class, which is the only class which should depend
on it. In particular the shared properties tests should be independent
Ok, and yes, that configuration change fixed at least one failure. The
additional comments are very welcome.
Don't run syncevo-http-server. It is not used by D-Bus testing.
Yes, this was cleared up in a previous mail.
Another possible difference might be how the script gets invoked. I
always run them in the out-of-tree build tree, inside "src". I run with
PATH=.:$PATH <absolute path to source>/test/dbus-test.py
If I've cleaned my system-wide directories of the syncevolution files
I find that this doesn't work. I've got to additionally add the
environment variables for the template and xml files. For me
With those things are fine... well, besides the failures we're trying to fix.
This copies <path>/test/dbus-test into ./dbus-test, which is
for some tests which expect a well-defined xdg root.
Your patch to use different names for the reference files and the local
copy makes sense. I don't normally notice such things because I never
run anything other than autogen in my source tree ;-}
Hopefully that can avoid a bit of confusion in the future for some poor soul.
In related news, I updated the test scripts so that D-Bus testing can
run as part of the regular nightly testing, in each of the chroots where
tests are run. I triggered a test run, but network problems broke it
because the SyncEvolution sources couldn't be updated from gitorious.
Otherwise I would have been able now to point to
and reports which contain D-Bus test
This would be very nice to have. Much less chance it'd fall into such
disrepair if it were run along side the other tests every night.