Re: Any Firefox / Opera WebRTC people here?

Closing the loop; we're now running the W3C WebRTC tests
<https://github.com/w3c/web-platform-tests/tree/master/mediacapture-streams>
as Chrome layout_tests. We do fail a bunch of tests in Chrome because we
don't conform, but now we'll at least not regress on the conformance we do
have :) We manage to run 7 tests cleanly, as you can see here
<http://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20%28dbg%29/builds/21663/steps/archive_webkit_tests_results/logs/stdio>
(search
for mediacapture-streams). However:

   - 10 tests are broken because we don't support video.srcObject = stream
   <http://crbug.com/387740> to assign streams.
   - We fail 46/61 cases in another one
   <https://code.google.com/p/chromium/issues/detail?id=405520&thanks=405520&ts=1408540572>
because
   the test and we don't agree what should go in a media stream track.

There are also some tests (like this one
<https://github.com/w3c/web-platform-tests/blob/master/mediacapture-streams/stream-api/mediastreamtrack/mediastreamtrack-end.html>)
that we can't run because they require permission denials. There's no way
we can do that in layout tests, so we can only run the ones that allow user
media.

James, Nils, how are you guys doing on the Firefox side?

/ P


On Mon, Jun 16, 2014 at 2:44 PM, James Graham <james@hoppipolla.co.uk>
wrote:

> On 16/06/14 13:20, Patrik Höglund wrote:
> > For the WebRTC media stream interop test
> > <https://webrtc-codereview.appspot.com/16719004/> I'm preparing, are
> there
> > any representatives from Firefox and Opera here who would be interested
> in
> > running the conformance tests? I just want to know who to talk to. It
> would
> > be great if we could all get the tests running continuously for all
> > browsers and raise awareness among our respective developers, so we all
> > become more standards compliant.
> >
> > I'm going to propose the test for review to web-platform-tests shortly.
> >
> > Don't worry if you don't pass all the tests; we certainly don't on the
> > Chrome side (42 pass, 28 fail, 3 timeout). ;)
>
> If those are just straightforward tests that don't require an actual
> WebRTC connection, they will be automatically be picked up by the test
> import machinery at Mozilla. Actually running the imported tests in CI
> depends on me ironing out a little instability in the results on OSX.
>
>
>