Timed Text Working Group Teleconference

<scribe> scribe: nigel

This meeting

nigel: Today we have some TTML topics to discuss and probably not much else...
... I have a non-update re TPAC.
... There's been some work this week on TTML, with a new pull request and some wiki
... discussion of time related matters.
... I should also raise the topic of Horizontal Review for TTML.
... Any other topics to discuss today, including AOB?

group: No other topics

TPAC 2016

action-475?

<trackbot> action-475 -- Nigel Megitt to Contact the chair of the web & tv ig to ask about schedule and joint meeting time. -- due 2016-07-28 -- OPEN

<trackbot> http://www.w3.org/AudioVideo/TT/tracker/actions/475

nigel: I did try to contact Yosuke of the Web & TV IG but he didn't yet reply.

tmichel: Yosuke is still around, I need to check the other chairs of the Web & TV IG.
... I'll send him a reminder.

nigel: Thanks!

tmichel: I'm trying to setup some kind of meeting for WebVTT to have it move forward at
... least to process the wide review comments that have been on hold for the last year.
... I hope by TPAC we can get an action plan.

glenn: I'm arriving on the Friday before the meeting starts so if anyone wants to get together
... in advance to talk about timing or anything we could arrange that.

nigel: I'm arriving on the Sunday late afternoon, so would be available.

tmichel: I will arrive on Sunday evening.

pal: I'll arrive on Sunday evening too.

glenn: [suggests organising a meeting over dinner on the Sunday evening]
... I'm staying at the Palacio de Governador.

nigel: I'm in the Mercure.
... Sounds like a good idea - let's try to organise that offline.

TTML1 & TTML2 issues, actions, PRs, editorial actions etc

action-462?

<trackbot> action-462 -- Glenn Adams to Create issue on ttml2 to add "mapping from other versions and profiles of ttml" -- due 2016-04-28 -- OPEN

<trackbot> http://www.w3.org/AudioVideo/TT/tracker/actions/462

glenn: That can wait a while longer.
... On the duplicate absolutize designators I have the input I need and can take that forward.
... On the br on timeContainer the real reason it can be a timeContainer is because it can
... have animation children.

nigel: I don't think that's the question - the PR just adds the BR element to the list.

glenn: I can go ahead and merge that PR too.

nigel: Great, is the next thing to talk about time?

glenn: Yeah... there's a lot there. In SMIL 3 §5.7.1 which covers terminology and concepts
... in SMIL there's a very appropriate short statement.

<gadams> SMIL3: Global time is defined relative to the common reference for all elements, the document root. This is sometimes also referred to as document time.

glenn: This is a key statement in my opinion in that it ties document time to the document
... root, which in our case is the tt element, which SMIL calls global time, which is time within
... the SMIL 3 context. I did a little historical research on TTML - if you go back to the
... TTML1 LCWD and before, we had begin and end time on the tt element itself, and we moved
... that between LC1 and LC2 over to the body element. And we altered some of the language
... in the definition of tt around that time. It's actually referenced in some of my comments
... in Nigel's walk-through.

nigel: To explain: I looked through TTML2 and SMIL3 to explain how the timing works and documented it in https://github.com/w3c/ttml2/wiki/Nigel's-TTML-and-SMIL-walkthrough
... Thanks Glenn for your comments. I think we're converging...

glenn: We didn't dive into this level of detail before. I think I agree that we're converging.
... My take on it is that the things that I thought were potentially broken for example in
... the current definition of document time and root temporal extent, I'm finding it's less
... broken than I thought it was. The confusion may have arisen around the use of the term
... implicit duration as opposed to active duration, in the text under the tt element.

pal: Do we really need to reference SMIL? Can we not define our own summary of timing semantics.
... It seems confusing to reference SMIL when the timing semantics can be succinctly expressed
... as in Nigel's summary. I don't think anyone will build a TTML implementation on a SMIL library.

glenn: I'm sympathetic to that.
... Writing new text and excising SMIL references would be a fair bit of work. Also, we
... may get it wrong and break something. On the other hand that may not be such a concern
... given the complexity of interpreting SMIL, which may lead to broken implementations
... in the first place. I'm not wedded to keeping the SMIL terms but it would take a fair bit of
... editorial work.

pal: I'm somewhat encouraged by the fact that the model that Nigel proposed is pretty
... intuitive and straightforward.

nigel: What I wrote is probably not algorithmically complete, but should have all the rules
... in that allow the algorithms to be derived.
... There is some discussion of SMIL in chrome, where there was a much argued deprecation notice.
... Google seems to be reconsidering that. It's in the context of animating SVG.
... We could non-normatively reference SMIL as the origin for the cut-down timing semantics.

pal: +1 to that.

glenn: This is also similar to the situation we have with style, and XSL-FO and CSS. I think that
... may even be more complicated than timing, but both are certainly complex.
... There was another attempt to do this by Sean, in our change documents. In his proposal
... for mapping to HTML and CSS he had an extremely simplified timing resolution model,
... which I decided at some point was inadequate.

http://www.w3.org/wiki/TTML/ChangeProposalIndex

glenn: Here's Sean's attempt to simplify the algorithm without referring to SMIL: https://github.com/w3c/ttml2/wiki/Nigel's-TTML-and-SMIL-walkthrough
... It would be interesting to compare this attempt to yours.

nigel: I haven't looked at that for a very very long time!

<gadams> https://github.com/skynav/ttt/blob/master/ttt-ttx/src/main/java/com/skynav/ttx/transformer/isd/TimingState.java

glenn: I pasted above the link to the main file in TTX that deals with timing state.
... It has functions like resolveActive() etc so this was our attempt to try to implement the SMIL semantics.

pal: I see the concepts of computing the active interval and positioning that active interval on an external timeline as being independent steps.
... It's like root container and layout.

glenn: That's right. They both have the concept of clipping.

nigel: +1

pal: Maybe we can first focus on unambiguously specifying what is the "effective" begin and end
... of every element on the document's abstract timeline and secondly what happens when
... you map it to that external context.

glenn: The notion of an active interval or duration in SMIL had to be different in SMIL because
... of the notion of repeat and min and max attributes, so it may be possible to jettison the
... concepts of active vs simple.

nigel: I think we do need to make sure that we unambiguously define which frames are
... dropped in drop modes. Also of course when using a clock timebase the relationship between
... document time and real time becomes somewhat more explicit. The two concepts can
... still be treated orthogonally.

group: some discussion of smpte mode discontinuous and conversion to continuous times, and the reference time for dropMode being 0.

nigel: We're out of time for today. It feels like we're making some headway on the wiki so I
... suggest we continue on there, and also each look at the other documents that have been
... referenced here today.

glenn: Sounds good.

pal: When I can access a computer I'll send a summary to the reflector of my really dumb
... mental model for how smpte discontinuous works.

nigel: Thanks all. [adjourns meeting]