Hey, folks. Looking over the OpenQA stuff in 
https://bitbucket.org/rajcze/openqa_fedora_tools/src/d48b884aa1244f5acbb9cad052f5b8a765cc08ad/tools/openqa_trigger/?at=master
 , I think it might be nice to use some of python-wikitcms' existing 
abilities and also extend python-wikitcms a bit to cover some of the 
stuff it does.

So far, I see:

in openqa_trigger.py , read_currents() could already be replaced by 
use of Wiki.current_compose - that's exactly what it's for.

I've just now added (haven't yet tested or pushed to master, but 
expect it soon :>) a Wiki.current_event property which should give a 
ValidationEvent (either ComposeEvent or NightlyEvent, depending) 
instance for the 'current' compose. The Event objects have some useful 
properties. They already handle image locations to a certain extent - 
see the imageurl property. I can quite easily extend this to what the 
OpenQA integration needs (finding images by arch) - I can do that 
tomorrow, or if someone wants to send a patch I can merge it. That 
should allow download_rawhide_iso to be a simple 
urlgrabber.urlgrab(event.bootiso(arch)) or something along those lines.

Since this is all in Python anyway, instead of calling out to relval 
in report_job_results.py , we could probably just use the python-
wikitcms bits. You can get the ValidationPage object for the 
'Installation' test type from the Event easily, then you can just do 
what relval does: produce a Result object for the result and call the 
ValidationPage's add_result() method for the appropriate ResultRow 
(you already have all the necessary info to find the appropriate row 
in conf_test_suites.py, though it should probably be improved to 
separate 'test case' from 'test name' - in the very first entry, the 
'test case' is actually "QA:Testcase_Boot_default_install" and the 
'test name' is "Server offline").

I did finish up the 'report-auto' stuff and cut a release today, so 
that's certainly available if we want to keep the idea of calling out 
to the relval command, but somehow it just feels like it'd maybe be a 
bit cleaner to use python-wikitcms directly.

Does this all sound reasonable? If no-one objects, I'll work on 
extending python-wikitcms' handling of images tomorrow, then submit 
some patches for the OpenQA bits.


-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net

_______________________________________________
qa-devel mailing list
qa-devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/qa-devel

Reply via email to