On Fri, May 16, 2014 at 10:35:48PM +0200, Paolo Bonzini wrote:
> Il 16/05/2014 22:20, Ademar Reis ha scritto:
> >>> * arm
> >>> * virt
> >>> - filter: "only hardware->disks->virtio"
> >>>
> >>> It would conflict with the "only hardware->disks" in the tests.
> >Filters are combined, so I don't see the conflict. "only
> >hardware->disks" is stronger than "only hardware->disks->virtio".
> >
> >The end result will be a tree without any disk variants.
>
> So arm is supposed to have
>
> drive_format = virtio
>
> to pick the default?
>
> This looks okay, but isn't it in some sense a layering violation? Why
> should an arm machine know what it takes to make tests run with a virtio
> disk?
The user can provide a different set of defaults to the test
runner, which will behave differently. Actually, if we end up
supporting ARM, we'll probably provide these defaults in the
framework.
The multiplexer just generates a combination of variables that
turns one test "into many tests" (one for each combination of
variants in the set), letting the decision to where and how to
run them to the test runner.
When running on ARM, a standard test runner could:
- Use a filter that chops out the incompatibilities before the
multiplexing happens (we could maintain an arm.filter file,
or just add an arm branch filled with filters).
- Skip the tests that can't be run because the test
requirements don't match what the platform is providing.
Hope it clears things up.
Thanks.
- Ademar
--
Ademar de Souza Reis Jr.
Red Hat
^[:wq!
_______________________________________________
Virt-test-devel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/virt-test-devel
