[PATCH weston] tests: Adding simple waycheck validation tool.

[Pekka Paalanen]

On Thu, 5 Nov 2015 12:32:02 -0500
"Jon A. Cruz" <jonc at osg.samsung.com> wrote:

> 
> 
> On 11/05/2015 10:17 AM, Pekka Paalanen wrote:
> > On Fri, 16 Oct 2015 12:23:47 -0700
> > "Jon A. Cruz" <jonc at osg.samsung.com> wrote:
> > 
> >> On 10/15/2015 01:51 AM, Pekka Paalanen wrote:
> >>> I think it's better to land this stuff in pieces than massage a
> >>> humongous replace-the-world patch series, if we can tell the pieces are
> >>> going in the right direction. The bits here are mostly just following
> >>> the existing weston-test-client-helper.c (which the commit message
> >>> forgot to mention).
> >>>
> >>> But yeah, probably makes sense to see how hooking even this stuff up to
> >>> 'make check' would happen before landing this. That will be one of the
> >>> most interesting things here. This patch as is does not add anything to
> >>> 'make check' which is a little awkward for a series adding new test
> >>> code.
> >>>
> >>> As for writing tests in the mean time, people should just ignore the
> >>> new framework for now, and write tests as if it didn't exist as long as
> >>> it doesn't support what the old framework does.
> >>>
> >>> This way we can incrementally advance on both fronts at the same time.
> >>
> >> So, the immediate question would now be as to how I should stage things.
> >> Should the changes that will go into 'make check' be a separate patch
> >> set or should it be an additional patch in this set?
> >>
> >> I'd say that running it as a separate set might make juggling git a
> >> little easier on me. But not to a degree to outweigh what would make
> >> reviewing things easier.
> > 
> > Hi Jon,
> > 
> > as I'm not sure how you are going to hook things up to Weston and it
> > will be interesting to see how you do it, I would like to have it as a
> > part of this series adding waycheck. In my view, you can't even
> > demonstrate one without the other: waycheck needs a compositor, and
> > seeing how 'make check' starts a compositor for a test needs the test
> > client.
> > 
> > I would very much like to see both pieces at the same time, in case one
> > requires changes in the other. It would be nice to keep in mind all the
> > various test styles (client, module, ...) we listed in the past, so
> > that adding a new case later shouldn't cause a huge refactoring.
> > 
> > By test styles I mean basically the different cases in
> > tests/weston-tests-env script.
> > 
> 
> One aspect for waycheck is that it could be run against most any
> compositor at any time, so allowing for those uses makes it less of a
> driving factor and more of a client.

Right. However my point is, if waycheck doesn't run as part of 'make
check', likely no-one will run it. This is my major issue with the
waycheck patch.


Thanks,
pq

> The other tests, however, do need to start weston and with different
> configurations. Those then fall into the potential white-box testing
> world, while waycheck is more for black-box validation.
> 
> 
> I've been going through some general cleanup, but will include the some
> of the current 'make check' tests conversions also. Might take a little
> bit more, but we definitely want to keep things easiest to review.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 811 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freedesktop.org/archives/wayland-devel/attachments/20151106/889a70bc/attachment.sig>