[Matroska-devel] FFV1/MKV IETF Session followup

Sebastian G. <bastik> bastik.public.mailinglist at gmx.de
Sat Jul 25 09:37:26 CEST 2015


25.07.2015, 06:48 Steve Lhomme:
> On Fri, Jul 24, 2015 at 3:36 PM, Sebastian G. <bastik>
> <bastik.public.mailinglist at gmx.de> wrote:
>> 23.07.2015, 01:44 Dave Rice:
>>> Hi all,
>>
>> Hello,
>>
>> if I have to summarize my reply it is 'subset'.
>>
>>> that develop the Matroska and FFV1 specifications should not simply
>>> throw the specifications "over the wall" into IETF, but to move
>>> specification development itself to an IETF context in a way that
>>> brings our discussions currently in ffmpeg-devel and matroska-devel
>>> and integrates them into an IETF context.
>>
>> I still wonder if the IETF welcomes it, anyway.
> 
> There's a good chance and it's worth trying. I presented Matroska once
> at the W3C and it didn't gain much traction. WebM wasn't much at that
> time and Opus not there yet. It would make sense now to do it at the
> IETF level which deals with lower level stuff than the web.

Low-level sounds good, because Matroska isn't just web releated.

>> How does one play a Matroska file with Firefox? It supports WebM which
>> is a subset of Matroska, but I am unable to play a Matroska file that
>> contains an h.264 encoded video, which is supported by Firefox as well.
> 
> That's software restriction. From what I know they decided (and I
> think Chrome too) to only deal wih WebM to avoid having to support all
> possible codecs. At that time Mozilla was refusing H.264. Now that
> they support it via GPU or Cisco, they should rethink their original
> choice.

Yes, I agree when it comes to Matroska and h.264.

>>>> Tessa said that although the specifications are already available
>>>> and the work is complete, the specifications would benefit from the
>>>> IETF review process.
>>
>> The specifications would probably benefit from any review process.
> 
> Contrary to popular belief, the format has been discussed and reviewed
> a lot (although considered as vaporware by many at the time, so not
> many people were actually involved). The specification text, not so
> much. And that's needed.

I was referring to specification text in general. People being involved
all the time might know some reasons and intentions behind a
specification text, but someone on the outside might misinterpret the
text and implement something that would violate the intentions of the specs.

>>>> Ben Campbell asked if it was ok if IETF take
>>>> over change control, and Tessa said that was understood. Ted Hardy
>>>> said that the community would need to participate in the IETF or
>>>> the effort would fail. Tessa said that she hoped the community and
>>>> IETF would come together. Steve Lhome, an original author of
>>>> Matroska, said he would continue to participate where ever the work
>>>> was going to happen.
>>
>> What would 'take control' actually mean? Assuming Matroska v4 gets
>> standardized, could a potential Matroska v5 be created without the IETF?
> 
> Not sure, but I think it may mean that discussions should happen at a
> ML at the IETF, some IETF meetings will take decisions on the format.
> I personally don't care where/how it happens as long as it's not
> inconvenient. The existing format will not go away. As for the future,
> we'll see if there is demand or not. There are a few things that could
> be done better in Matroska and if in the end there's a derivative
> format designed at the IETF, why not. Even if it breaks backward
> compatibility. Bringing that new format means adding the old one
> should be easy too.

Sounds alright, I guess.

>> Speaking of subsets. The WebM Project takes Matroska and transforms it
>> into something new (a subset), but Matroska stays what it is.
>>
>> Why doesn't PREFORMA use a subset of Matroska, which would be
>> standardized on its own?
> 
> We'll see how the discussions go. There's a chance some features are
> left out. But it's OK to make different RFCs as additions on top of
> the core RFC.
> 

Best Regards,
Sebastian G. (bastik)




More information about the Matroska-devel mailing list