[webkit-gtk] Question about Media Source Extension

Cédric Bellegarde cedric.bellegarde at adishatz.org
Wed Feb 28 05:31:35 PST 2018


My bad, after reading Philippe comment, I was searching how the hell 
was this possible. And after testing with epiphany, it was more: "how 
the hell is this possible in Eolie?".

And the bug was in Eolie, one line was setting it off and later another 
one was setting it on... Sorry for the noise.

BTW, I will let it on in Eolie as this may allow you to get more 
backtrace.

I'm trying to reproduce one crash for now with: full bt + gstreamer 
debug files.

regards,

Le mer. 28 févr. 2018 à 14:24, Michael Catanzaro 
<mcatanzaro at igalia.com> a écrit :
> On Wed, Feb 28, 2018 at 1:20 AM, Cédric Bellegarde 
> <cedric.bellegarde at adishatz.org> wrote:
>> And here my surprise, with -DENABLE_MEDIA_SOURCE=ON and 
>> enable-mediasource=False, 1080p is available in YouTube and 
>> WebKitGTK does not crash anymore.
> 
> WOW! This is a major discovery.
> 
> The good news is that you must have been testing our stable release 
> rather than  our latest development release, WebKitGTK+ 2.19.91, 
> where we actually have enabled the ENABLE_MEDIA_SOURCE=ON build 
> option. It's still off by default at runtime, because it does not 
> work reliably yet, but that is exactly the configuration that you've 
> requested.
> 
> It seems quite bad that enabling the build option but not the runtime 
> option enables higher-resolution video playback on YouTube, though, 
> because that means that the feature is somehow web-exposed such that 
> YouTube can detect it, even though it is supposed to be disabled at 
> runtime. That seems like a major bug. It indicates that the runtime 
> option does not function properly in some way or another, and that 
> leaving the build option enabled is risky, and that we should 
> reevaluate our choice to enable the build option. This is somewhat 
> urgent, because we have a new stable release coming up in a week and 
> a half. Enrique, Alicia, could you investigate this to see what is 
> wrong, please?
> 
> In the past, when the MSE feature was originally developed, WebKit 
> did not support hiding JavaScript features at runtime. That's no 
> longer the case -- nowadays, we would expect the MSE features to be 
> completely invisible to webpages when the runtime option is disabled 
> -- but perhaps that was never properly implemented for MSE. (That is 
> actually the reason why we have started using more runtime options in 
> WebKit, in preference to build options.) Anyway, I don't know how to 
> fix this; it requires some investigation.
> 
> Michael
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-gtk/attachments/20180228/6e12d6a7/attachment.html>


More information about the webkit-gtk mailing list