W3C home > Mailing lists > Public > public-html-testsuite@w3.org > June 2010

Re: Canvas, Audio, Video Test Submission

From: Matthew Gregan <kinetik@flim.org>
Date: Tue, 29 Jun 2010 12:15:27 +1200
To: Kris Krueger <krisk@microsoft.com>
Cc: James Graham <jgraham@opera.com>, Simon Pieters <simonp@opera.com>, "'public-html-testsuite@w3.org' (public-html-testsuite@w3.org)" <public-html-testsuite@w3.org>
Message-ID: <20100629001527.GI29988@flim.org>
Hi Kris,

common/video.js is still using UA sniffing in getVideoExtension, which
causes the first test (video.html) to fail on Firefox nightly builds.

common/media.js is also still using UA sniffing in getVideoType, but I can't
see that being used anywhere in the current tests.

The other tests now behave as expected.  Thanks for fixing this!

Cheers,
-mjg

At 2010-06-28T23:41:51+0000, Kris Krueger wrote:
> Updated to use canPlayType - if you have other feedback, please speak up.
> 
> -Thx
> 
> -----Original Message-----
> From: public-html-testsuite-request@w3.org [mailto:public-html-testsuite-request@w3.org] On Behalf Of Kris Krueger
> Sent: Friday, June 25, 2010 4:19 PM
> To: James Graham
> Cc: Simon Pieters; Matthew Gregan; 'public-html-testsuite@w3.org' (public-html-testsuite@w3.org)
> Subject: RE: Canvas, Audio, Video Test Submission
> 
> OK - I'll take a peek at making this change, in theory everyone is properly returning null when they don't support a codec.
> 
> Though this is really a bad api design...
> If one implemented malloc that maybe returned a valid pointer or probably returned a valid pointer.
> The consumers of this malloc would be doomed to failure.
> 
> -Kris
> 
> -----Original Message-----
> From: James Graham [mailto:jgraham@opera.com] 
> Sent: Friday, June 25, 2010 9:18 AM
> To: Kris Krueger
> Cc: Simon Pieters; Matthew Gregan; 'public-html-testsuite@w3.org' (public-html-testsuite@w3.org)
> Subject: RE: Canvas, Audio, Video Test Submission
> 
> 
> 
> On Fri, 25 Jun 2010, Kris Krueger wrote:
> 
> > I understand that though the reason we didn't choose to use this was 
> > that the canPlayType call may not actually lead to media being able to 
> > be played. For example when a UA returns probably it doesn't actually 
> > mean that the media will play.
> >
> > If we want to use this then I would encourage to have this api changed 
> > to be binary (yes/no) so that a web dev can be confident that an end 
> > user will actually see media playing.
> 
> The reason that the API isn't like this is that the UA might not know which types it can play without just trying to play them. Nevertheless the problems with having UA sniffing in tests are far greater than the problem that a UA might think it is able to play a type that it cannot.
> 
> 
> 

-- 
Matthew Gregan                     |/
                                  /|                    kinetik@flim.org
Received on Wednesday, 30 June 2010 09:34:40 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 30 June 2010 09:34:43 GMT