On Mon, Sep 11, 2023 at 08:35:37AM -0400, N?colas F. R. A. Prado wrote:
> On Sat, Sep 09, 2023 at 12:08:22AM +0100, Mark Brown wrote:
> > Why?
> To better reflect the actual test plan status. If 0 tests were run, it doesn't
> really make sense to say that the test plan passed, rather it was skipped since
> nothing was run. So with this change, if there's a regression that prevents the
> soundcard driver from even probing, the result won't be "pass", but "skip", and
> the reason 'No soundcard available' will be in the logs.
So, I would interpret the overall result for the suite as being "No
errors were found in any of the cards discovered" if there is no
configuration file specified which enumerates the set of cards that are
expected (if there is a config file that's a different matter, we know
what we're expecting). I'm not sure that the different behaviour for 0
cards is super useful.
On Mon, Sep 11, 2023 at 03:29:00PM +0100, Mark Brown wrote:
> On Mon, Sep 11, 2023 at 08:35:37AM -0400, N?colas F. R. A. Prado wrote:
> > On Sat, Sep 09, 2023 at 12:08:22AM +0100, Mark Brown wrote:
>
> > > Why?
>
> > To better reflect the actual test plan status. If 0 tests were run, it doesn't
> > really make sense to say that the test plan passed, rather it was skipped since
> > nothing was run. So with this change, if there's a regression that prevents the
> > soundcard driver from even probing, the result won't be "pass", but "skip", and
> > the reason 'No soundcard available' will be in the logs.
>
> So, I would interpret the overall result for the suite as being "No
> errors were found in any of the cards discovered" if there is no
> configuration file specified which enumerates the set of cards that are
> expected (if there is a config file that's a different matter, we know
> what we're expecting). I'm not sure that the different behaviour for 0
> cards is super useful.
Right... So what we want to be doing is adding a config file for every platform
defining the card(s) and PCMs expected, so that when they're missing a test
failure will be triggered which is even more helpful. Although I've noticed that
only missing PCMs are detected currently, but I imagine it should be possible to
to extend the code to detect missing cards as well.
I take it the intention is to expand the conf.d directory with configs for all
platforms currently being tested then? There's only one example file there so I
wasn't sure.
Thanks,
N?colas
On Tue, Sep 12, 2023 at 03:23:34PM -0400, N?colas F. R. A. Prado wrote:
> On Mon, Sep 11, 2023 at 03:29:00PM +0100, Mark Brown wrote:
> > So, I would interpret the overall result for the suite as being "No
> > errors were found in any of the cards discovered" if there is no
> > configuration file specified which enumerates the set of cards that are
> > expected (if there is a config file that's a different matter, we know
> > what we're expecting). I'm not sure that the different behaviour for 0
> > cards is super useful.
> Right... So what we want to be doing is adding a config file for every platform
> defining the card(s) and PCMs expected, so that when they're missing a test
> failure will be triggered which is even more helpful. Although I've noticed that
> only missing PCMs are detected currently, but I imagine it should be possible to
> to extend the code to detect missing cards as well.
It seems like a reasonable approach for systems that do want to have
this confirmation.
> I take it the intention is to expand the conf.d directory with configs for all
> platforms currently being tested then? There's only one example file there so I
> wasn't sure.
I think it's a question for people working on individual systems if they
want that coverage, for example I don't really care for the things in my
CI because I have higher level stuff which will notice any newly missing
tests so I don't need the test to do anything here.