On 2014-06-04 10:36 AM, Gary Chen wrote:
> Wow~~ sounds pretty cool~
It does, doesn't it?
To be clear, when I say "we" I mean "no one one the gecko team". I don't
know who, if anyone, is actually working on this. Sounds like there are
partner requirements for it, so hopefully it's on someone's roadmap.
I don't see a gaia bug. There's some discussion in
https://bugzilla.mozilla.org/show_bug.cgi?id=941351 which is about
supporting HLS m3u8 directly as the source of an HTML video element,
like Apple Safari does.
> What is native implementation? Can I ask for more detail?
Native implementation would be adding an mpeg-ts parser in C++ to gecko
(perhaps the one from android) so video muxed this way can be accepted
directly by the MediaSource Extensions API. We don't want to do this if
we can avoid it. mpeg-ts is a legacy format, and C++ parsers add to our
security and maintenance surface.
The m3u8 parsing should still be done in js: there's no performance
advantage to C++ for that. If we decided to support HLS directly we
would add the same code to gecko as a jsm.
-r