Re: [Moq] One last question about relays and rate adaptation from our pre-lunch/PST discussion

There can be other functions beyond what you said below Luke. So, thats why
i am suggesting to use a simple naming (passive vs. active) where passive
is more or less what you describe as pass thru. Active ones are obviously
more involved and can be dealt with later.

On Fri, Feb 3, 2023 at 12:41 Luke Curley <kixelated@gmail.com> wrote:

> I agree with Ali.
>
> In my mental a "relay" is any node that sits between the broadcaster and
> viewer. But maybe "intermediary" is the better term for that.
>
>
> IMO a relay could: *passthough*, *transmux*, or *transcode*.
>
> A *passthough relay* forwards catalogs and objects, unable to decode
> them. This is what some people call a "relay", restricted to networking
> operations only.
>
> A *transmux relay* can modify the catalog but not objects. This is
> Christian's example of a relay that joins tracks from different sources,
> like captions, translations or even ads.
>
> A *transcode relay* can modify the catalog and modify objects. This could
> be a full re-encode or just inserting metadata into an existing bitstream.
>
> On Fri, Feb 3, 2023, 11:41 AM Ali C. Begen <ali.begen=
> 40networked.media@dmarc.ietf.org> wrote:
>
>>
>>
>> > On Feb 3, 2023, at 10:34 AM, Roberto Peon <fenix=
>> 40meta.com@dmarc.ietf.org> wrote:
>> >
>> > Relays clearly cannot do transcoding if they are required to function
>> without being able to access the bitstream plaintext(s).
>>
>> This question came up several times and looks like people still have the
>> (more or less) same ideas they had at the beginning. Saying relays cannot
>> mess with the media tracks since they won’t have access to clear media
>> simplifies things quite a bit but I see the reasons for those who would
>> like to override this.
>>
>> Maybe as we are defining the terminology for MOQ, we should separate the
>> two cases:
>> 1) passive relays: that are simple store/forward cache type of
>> intermediaries.
>> 2) active relays: these can do more sophisticated things (like
>> transcoding, merging, stitching, etc.) if certain requirements are
>> satisfied (we can debate what these are)
>>
>> This type of distinction may conclude this discussion and will hopefully
>> make everyone happy?
>>
>> -acbegen
>> --
>> Moq mailing list
>> Moq@ietf.org
>> https://www.ietf.org/mailman/listinfo/moq
>>
> --
-acbegen
Using iThumbs