----- "Alexander Todorov" <atodorov(a)redhat.com> wrote:
> Some questions:
> How do I test my changes before sending them as paches?
By installing autotest+autoqa and trying if it works. Currently you
need both autotest server and client to test it. Maybe if Will accepts
patch in ticket #52 it could help, because then you could just
install only the autotest client:
----- "James Laska" <jlaska(a)redhat.com> wrote:
> This topic came up elsewhere last week, how do you test your tests?
> current approach is to say ... just run your tests. This works
> but perhaps we can also offer testing in a 'autotest' environment. I
> think Kamil's ticket now makes sense to me.
> Kamil ... is there any documentation needed to help setup the
> environment as you have done in ticket#52?
I don't know if I exactly got what you mean, but after applying
the patch you can just edit /etc/autoqa.conf, set "run_locally =
True" and then you can test watchers/tests locally. For example like
$ autoqa post-repo-update -t repoclosure --arch x86_64 \
--name rawhide http://mirror/pub/fedora/linux/development/%a/os
13:40:20 INFO | Writing results to /usr/share/autotest/client/results/default
13:40:21 INFO | Initializing the state engine
13:40:21 INFO | Symlinking init scripts
13:40:26 INFO | START ---- ---- timestamp=1256560826 localtime=Oct 26 13:40:26
13:40:26 INFO | START repoclosure repoclosure timestamp=1256560826 localtime=Oct 26 13:40:26
13:41:01 INFO | Test started. Number of iterations: 1
13:41:01 INFO | Executing iteration 1 of 1
13:43:26 INFO | Test finished after 1 iterations.
13:43:27 INFO | GOOD repoclosure repoclosure timestamp=1256561007 localtime=Oct 26 13:43:27 completed successfully
13:43:27 INFO | END GOOD repoclosure repoclosure timestamp=1256561007 localtime=Oct 26 13:43:27
13:43:28 INFO | END GOOD ---- ---- timestamp=1256561008 localtime=Oct 26 13:43:28
Currently you can't do that without autotest_server install and
configuring at least one client (which will be probably localhost
anyway, so why all the bother?).
I guess if the patch is applied (it was a fast fix, I suppose wwoods
may tweak according to his views), it would be nice to tell people
"Hey, you don't have to install autotest_server if you don't want,
just edit autoqa.conf and test your script locally". This can be added
to the wiki or to README.
here's my first take at this test case.
It is based on the control file in post-koji-build hook. As Will said I assume
that the baseurl parameter points to the rpm file which I wget and run rpmlint
How do I test my changes before sending them as paches?
Do I need to add this test case name to post-koji-build/testcases ?
Please review the attached file. Comments are welcome.
I have a test which runs rpmlint against all packages in the distro and I'd like
to convert it to AutoQA format. I've read
https://fedoraproject.org/wiki/Writing_AutoQA_Tests but it doesn't mention what
parameters are available to tests to use. My original test accepts NFS path,
mounts it, finds all *.rpm files and runs rpmlint -i against all of them. Is it
possible to access the storage media from the test case or to have a pointer to
the rpm files itself (if running against a particular package)?