[Piglit] Crashing tests are not shown in the HTML summary
Dylan Baker
baker.dylan.c at gmail.com
Mon Apr 7 22:16:23 PDT 2014
On Monday, April 07, 2014 10:48:10 PM Ilia Mirkin wrote:
> On Mon, Apr 7, 2014 at 7:49 PM, Dylan Baker <baker.dylan.c at gmail.com> wrote:
> > Actually, branches do have a pass fail status, it's the worst status of
> > that branch's children.
> >
> > I've been thinking of using a Tree structure to represent the tests in
> > Summary. The problem with the current architecture (as it pertains to this
> > problem) is that the the statuses of each group are auto-generated from
> > the
> > 'worst' status in their children, and the only way to set a group status
> > is
> > after the entire tree has been generated.
> >
> > If you write a patch to add a fake subtest I'll be sure to look at it.
>
> It's actually worse than I thought. It's not a problem generating the
> summary, it's a problem generating the results file in the first
> place.
>
> This logic is in Test.execute:
>
> if 'subtest' in result and len(result['subtest']) > 1:
> for test in result['subtest']:
> result['result'] = result['subtest'][test]
> json_writer.write_dict_item(os.path.join(path,
> test), result)
If there is only one subtest the subtest is written out and it's presented in
summary with the subtest status as the test status. at least, that was the
reasoning.
>
> (why is it > 1 and not > 0? whatever.) In any case, this ensures that
> we don't even have the right stuff in the result, so no amount of code
> changes to summary can fix it. I thought of throwing in a comparison
> of the statuses, but realized they're strings. Would it be an API
> abuse to import framework.status and parse them here?
I don't think so. I did that originally, but decided that since we didn't need
to there wasn't a lot of reason to do a worthless string > object conversion.
>
> -ilia
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.freedesktop.org/archives/piglit/attachments/20140407/69e97737/attachment.sig>
More information about the Piglit
mailing list