Timed Text Working Group Teleconference

22 Oct 2018

See also: IRC log


Glenn, Nigel, Andreas, David_Singer, Pierre


<scribe> scribe: nigel

Agenda bash

Nigel: Looking at our agenda:

TTWG TPAC Agenda page

Nigel: I don't think we will have a joint meeting with CSS WG

Andreas: Tomorrow I'd like to discuss closer alignment with CSS

Glenn: I'd like to consider TTML2 2nd Ed requirements

Nigel: We have an agenda topic for that tomorrow - I have a use case to bring also.

Andreas: For the agenda, I don't think we need anything for TTML <--> WebVTT mapping.

Nigel: Yes, we took that off the agenda already.
... Glenn asked to discuss Karaoke.

Glenn: I don't have anything prepared for that, so we can scratch that off the agenda
... or cover it very quickly.

Andreas: There will be topics relative to the TTWG that will come up during the M&E IG,
... like 360º VR and AR.
... I have to be at the AC tomorrow from 3 so I will not be able to join the future requirements
... session.
... Can we move the future requirements to after lunch and then audio profiles later?

Nigel: OK

Andreas: I need to go to the M&E IG now, see you there later!

<glenn> Note: github.com is undergoing technical problems, see https://status.github.com/messages

Nigel: I will start to prepare some slides for our joint meeting with M&E IG now.

Glenn: I think we need to cover WebVTT and the status

Pierre: We should discuss the TTML ballots also
... If it helps we could walk people through the implementations

Nigel: When should we cover WebVTT?

David: My feeling now is that we won't get enough input to get to PR and Rec so we
... should leave it there and propose that we drop the work and publish as a WG Note.
... If nobody comes forward to say "I'm willing to do the work" then we should go with
... that proposal.

Nigel: Do we need an agenda topic to take that any further?

David: I will check if anyone else from Apple is likely to attend today or tomorrow.

Nigel: The main body of our meeting is tomorrow - I have prepared some material for
... the discussion on live subtitle topic.

Pierre: If the group decides to tackle this, the first step should be for the WG to accept
... the system model, as the basis for the work.

Nigel: +1

Pierre: I think there's real confusion about the production process vs the "broadcast encoder"
... for want of a better term. Once it hits ISOBMFF, what happens, for example? Does
... anything special have to happen after the emission encoder or is it really upstream.
... Even if we could publish a Note and tell people to look for it before asking questions
... that would be helpful.

Nigel: I feel I will be able to answer that question!

David: On the WebVTT proposal, I will take an action to send a note to the mailing list
... explaining the proposal. If we get no response we should take silence as assent.

M&E IG joint meeting preparation

Nigel: [bashes presentation deck for joint M&E IG]

Pierre: For future requirements for TTML, it would be useful to think about the language
... in the core about style attributes, and allowing for additional modules.

Glenn: If we have modules that try to apply diffs to the core, that becomes messy.

Pierre: Maybe we would choose not to support that, but to support plugins for example

<scribe> .. new style attributes that participate in the style resolution algorithm.

Glenn: I don't know if anyone wants to discuss 3D and adding a z-value dimension to
... text positioning.
... I'm having a conversation about future requirements concerning a DOM API
... for TTML, from an organisation interested in translation and localisation.
... It would be to support Javascript processing of TTML for example.
... Perhaps that's the conceptual model that the organisation I'm talking to has in mind.

Nigel: I've uploaded that to https://www.w3.org/2011/webtv/wiki/Face-to-face_meeting_during_TPAC_2018#14:45_-_15:30_Timed_Text_Working_Group_joint_meeting

TTML Profile Registry issues

Nigel: Given that GitHub is down, this may not work too well...
... I can see we have some new issues.

Glenn: Nigel had a question about #52
... It's a portmanteau issue where the details will be in the Pull Request

Nigel: Please don't do that - you're asserting there's something wrong with the text
... and planning to go ahead with a pull request before getting any discussion that
... concurs with your review. Be prepared to get push back from the WG on that.

Pierre: But what is wrong with the text?

Glenn: For example Respec complains about a normative reference in a non-normative section.
... Another is inconsistent capitalisation.

Pierre: OK, editorial things.

Glenn: I'll throw them in the summary to help with review.

Pierre: Are there any non-editorial issues?

Glenn: If there are then I will open a separate issue for them.
... There's also some language in the prologue to the media registration part that talks
... about plans for sending to IANA but we have already done that so it no longer applies.
... It can either be removed or truncated.
... The biggest change I'm going to propose is changing §4.2 to informative. I'm not sure
... how it got in there, I wasn't paying attention apparently. It's quite problematic.
... It purports to describe how to resolve the profile of a document.

Pierre: That was before TTML2
... We should just fix it.

Glenn: I'm willing to keep it in as an informative section.

Nigel: It precedes TTML2 and the referenced specifications describe their own ways of
... identifying the document instance's profile. In that sense it is informative because it's
... a gathering of other information.

Pierre: One possibility is to remove §4.2 - that would be my preference.

Glenn: That would be my preference too.

Pierre: This is trying to be helpful, I understand.

David: It seems like a good answer to help people derive what profile a TTML document instance is.
... It should be informative.

Glenn: The trouble is that it is not what TTML1 or TTML2 says.

David: It shouldn't be wrong!

Pierre: Right, it is possible to summarise, and should not be too much trouble, to say
... where to look at in a TTML document instance to find what kind of thing it is.

Glenn: I don't even mind suggesting one possibility of using XPath. Maintaining this
... data is not part of the registry.

Pierre: And its wrong!

Nigel: Is it?

Pierre: Yes, especially in TTML2 you have precedence rules about content profile, profile
... attribute...

Glenn: profile override, outer context...

Pierre: yeah

Glenn: For example for tt1t it doesn't refer to what TTML1 or TTML2 does.

Nigel: The XPath thing doesn't say you must use XPath.

Pierre: IMSC does not use this technique for deciding if a document is IMSC compatible for instance.

Glenn: I don't know of any implementation that uses this technique.

Pierre: I would replace it with a more descriptive prose section.

David: It is useful to help find what kind of TTML document instance you have in front of you.
... What the document claims to be.

Pierre: You can validate any document against any profile to see if it valid.

David: You need to know what a document claims conformance to.

Glenn: You're conflating content profile vs processor profile.

Nigel: I didn't hear that from David.

David: It seems like a reasonable question to ask what a document claims conformance to.

Glenn: It's out of place in the registry because the codecs profile does not refer to content profiles.

Nigel: That's true.
... I would maybe rewrite it and put it in an appendix rather than have it in §4.2.

David: The codecs parameter is useful and you need to know how to specify it based on a document

Nigel: Subtle distinction: it's about knowledge of the document, not the document itself.
... It may be that you know a doc is EBU-TT-D and IMSC but the document doesn't make
... any claim internally. You can set codecs based on that.

Glenn: You also may know you don't need full processor support for every feature.
... By the way the document doesn't set a precedence order for profile vs codecs

David: Why is it called codecs?

Pierre: So it can be used in MP4

David: You can copy the codecs value across?

Pierre: Yes

<glenn> github outage incident report: https://blog.github.com/2018-10-21-october21-incident-report/

Nigel: This is what goes after stpp.

David: It's the sub-parameter for the codecs? Got it.

Glenn: My action is I plan to make a PR for this based on the input I've heard today.
... Instead of removing it entirely I'll take the approach that Pierre suggests, to try to
... summarise it, and add a statement that it's a non-normative section.

Nigel: I would consider moving it to an appendix too.

Glenn: That would be my preference too. I may leave behind a pointer.

Nigel: I don't think that's needed.
... It would be useful to merge #43 before you do a pull for #53 Glenn

Glenn: I agree, please can we merge them soon?

Nigel: Any objections to merging the open pull requests?

Glenn: We will probably end up tweaking the result of that to deal with other issues.

Nigel: I'm not hearing any objections.
... Please note Mike is the editor here and wants to continue with that.

Pierre: Just create a pull request and Mike can be in charge of merging it.

Nigel: Will you do the pull to add TTML2 profiles Glenn?

Glenn: Yes, and for #38 (Content Profiles vs Processor Profiles) too.

Nigel: That's all on this topic. I propose we take a break and then join the M&E IG for
... the rest of the day.

Glenn: I'm going to join that later.

Nigel: Okay!

Break, then Media and Entertainment IG.

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version 1.154 (CVS log)
$Date: 2018/10/22 14:29:40 $