Meeting minutes
Agenda
Chairing update
ChrisN: Welcome Chris
Lorenzo from Comcast as new MEIG co-Chair
… Thank your Pierre for his
contributions
ChrisL: (gives
introductions)
… Excited to join W3C, thanks for having me
here!
ChrisN: It's great to
have you here, look forward to working with you.
… Also I've become Media WG co-Chair in addition to
MEIG.
… Hopefully can use this to build relations between
the two groups, and external groups.
… I want to mention two workshops Pierre is
involved:
… wide color gamut and high dynamic range for the
web (as well as the Community Goup on this topic)
… and professional media production on the
web
… More detail and workshop dates are
tbd.
Rechartering
ChrisN: AC review is
ongoing and closes on May-28
… 9 responses so far
… Thank you to those who voted so far. If you
haven't voted yet, please ask your AC rep.
… It would be helpful to have as many votes as
possible
Multicast receiver API
ChrisN: We discussed
this during the previous call, with Jake Holland from Akamai.
… There's number of open questions there: threat
model (e.g, based on the WebRTC threat model),
… whether to have a WebTransport like API or
use-case specific APIs.
… We had a followup meeting with Jake, the Chairs
and the Team on Apr 30.
… The main question to gather people interested for
use cases and technical discussions.
… It seems to me there is interest in media use
cases for live media streaming and large file
downloads.
… What we plan to do is work with Jake to create a
dedicated CG for further discussion.
… Any thoughts or questions?
<kazho> Actually some Japanese Broadcasters have the sort of idea. I'll talk to them.
Igarashi: would like to
understand the use cases to achieve this proposed API,
… which part to be standardized by IETF,
etc.
ChrisN: It seems that
most of the work is related to the protocol level, so IETF.
… And there are two parts for us at
W3C:
… 1. Understand the use cases, security concerns,
etc.
… 2. What the API shape should be based on use
cases: WebTransport like or high level to integrate with
<video> element playback, for example.
Igarashi: One of the
concerns might be other kind of deployment issue for multicast,
e.g., due to support in home routers.
… Most routers in Japan don't support multicast
capability,
… so wanted to ask which part to be handled by
IETF.
ChrisN: It's a good
question. It's not my area of expertise,
… but once the potential CG is created, that is a
good place to have the discussion about that as well.
Igarashi: Also we
should look into the applicability,
… for example, wondering about the Comcast router
in US
ChrisN: Thanks, that is
useful input.
… The other aspect is that Jake is kind of new to
W3C, and so we'll give help from our side
Igarashi: Do you know any approach related to multicast in US?
ChrisL: Could reach out
people inside Comcast.
… Issues on streaming and network congestion,
etc.
ChrisN: Yes, multicast is enabled in some provider networks in UK.
Igarashi: Some of the JP vendors provide a bit different kind of router for that purpose.
ChrisN: I guess one of the points of this work is to make it more generally available on the internet.
kaz: We should look at use case and requirements, also see which parts of the work belong in W3C and which in IETF.
Media Timed Events
ChrisN: In our work on
Media Timed Events, we've had discussion with TTWG on unbounded
cues in WebVTT.
… We would like to gather technical use cases and
requirements.
… We made the API change to support this in
TextTrackCue in HTML.
… Some questions have been raised: a WebVTT syntax
for unbounded cues to be defined.
… The WebVMT format proposed by Rob has some
syntax.
… Also processing model for updating the unbounded
cues.
… Suggestion from the TTWG was having concrete
technical use cases
… to enable further discussion and design of
possible solutions based on good understanding of the
requirements
<cpn>
https://
ChrisN: I've created a
document to capture use cases and requirements.
… I would like to get your initial feedback on the
document itself.
… It starts with links to the background
discussion, including the discussion with the Timed Text
group.
… It then consolidates the use cases. Please let me
know if anything is missing.
… The use cases includes one about sports score
updates.
… The cue value could be a JSON
object.
… The score starts at "0" and gets updated at
different times during the video.
… The 2nd use case is about subtitles, which has
two sub-cases:
… a. segmented WebVTT files, and b. WebVTT carried
in fragmented MP4
… My question here is if those use cases are the
right ones.
… Also any other use cases to be added?
Nigel: It's great to
see this.
… The main point is if the use cases make the
changes which require implementation changes of WebVTT
… Updating the cue is possible from JS, but not
from the WebVTT document. There's a big difference between the API
and what can you do in a VTT file.
… The use cases could be tightly scoped to updating
end time of unbounded cue. It would help to have the use cases, but
also consider a wider set of use cases.
… Now's a good time to raise any and other
requirements. We want to avoid going down a cul-de-sac route by
limiting scope.
… Implementers have been resistant to making
changes, so we should consider that point when we generate use
cases.
ChrisN: Thanks
Rob: Unbounded cues are a very common use case in WebVMT
Rob: Problem: how to
represent in WebVTT?
… If you want to create a cue, do so without end
time, common in live media.
… Sports score is an example. I proposed that as a
text example, rather than JSON. Update it during a live
event.
… Don't know how long the cue will need to be, as
it's a live event.
… Proposals in WebVTT issue https://
… I can see two families of use cases: one is like
chapters, where it starts but you don't know how long it
lasts.
… The other is a daisy-chain where you have a
metadata or text value, and the value is updated repeatedly during
the duration of the media.
… The subtitle containing the score persists
throughout the match, then it updates.
… Interested in feedback in other use
cases.
… To Nigel's point on other changes that could be
made to link the VTT to the JS API, I'm interested to know what
those are.
… The JS API allows many things to change. The
current changes are limited in scope.
… Creating an unbounded cue and later setting it it
equivalent to the VTTCue constructor.
ChrisN: I had a similar
question. Is parity with the JS API a design goal for the WebVTT
change?
… JS allows you to change any aspect of a cue at
any time.
… Also how to handle it in a backwards-compatible
way?
… I don't think we have all the answers today, so
need follow on discussion.
<nigel> +1 to it being a useful starting point!
ChrisN: Thinking about the use cases and requirements should be a useful starting point (shows the potential requirements)
Rob: Yes, it's a useful exercise
ChrisN: what I'm
thinking is to turn this into a markdown document
… and put it into a GitHub repo so that we can
update it using GitHub pull requests.
… To do that, we need to decide where to put
it.
… Happy to use our MEIG repo, and then people are
encouraged to make contributions.
… My own time to contribte to the document is very
limited, so I'll be relying on all of you to help.
… Does that sound like a reasonable
approach?
<nigel> +1 to MD in MEIG repo
<RobSmith> +1
kaz: +1
ChrisN: This is part of
the MTE discussion, the next meeting of the MTE TF will be on
Monday, May 17.
… I will put the document into MEIG's GitHub
repo,
… and let's add to the use case
descriptions.
AOB and next MEIG call
ChrisN: Any other business for today?
(none)
ChrisN: Our next MEIG
meeting will be June 1. That's 3 weeks away.
… No specific topic yet, please let me know if you
have any suggestions
[adjourned]