Timed Text Working Group Teleconference

<trackbot> Date: 18 December 2014

<scribe> scribeNick: nigel

This meeting

nigel: introduces agenda
... Is there any other business?

pal: I'd like to review the TTML2 timeline and the IMSC 1 timeline if there's time at the end, and perhaps
... start to think about IMSC 2 timeline.
... and how long it takes to complete PR review, i.e. PR -> Rec
... This is because there are other organisations that are considering using those specs.

Action Items

action-359?

<trackbot> action-359 -- Thierry Michel to Draft a message re imsc 1 cr test suite and implementation and a list of bodies to send it to. -- due 2014-12-18 -- OPEN

<trackbot> http://www.w3.org/AudioVideo/TT/tracker/actions/359

tmichel: I sent the message, and copied to the member-tt list and I sent it to all the people that are listed
... on our charter as liaison and the others to whom Nigel sent the wide review messages, and two
... coordination groups.
... I copied pal, plh, nigel and dsinger.

close action-359

<trackbot> Closed action-359.

action-360?

<trackbot> action-360 -- Nigel Megitt to Send liaison request on behalf of w3c ttwg to arib -- due 2014-12-18 -- OPEN

<trackbot> http://www.w3.org/AudioVideo/TT/tracker/actions/360

nigel: I have begun drafting this and have addresses for the recipients - they have requested a paper
... document requesting the liaison. I have more to do on this.

Issues

issue-21?

<trackbot> issue-21 -- window anchor points not supported? -- pending review

<trackbot> http://www.w3.org/AudioVideo/TT/tracker/issues/21

glenn: I haven't received any comments on any of the issues that I've marked as Pending Review
... I guess most likely people haven't had a chance to dive into it. I suggest we close this and if an issue
... arises then it can be filed.

close issue-21

<trackbot> Closed issue-21.

issue-236?

<trackbot> issue-236 -- Character spacing, i.e. letter-spacing -- pending review

<trackbot> http://www.w3.org/AudioVideo/TT/tracker/issues/236

glenn: this is based on the CSS attribute of the same name, so should be quite straightforward.
... This was to define a property to control tracking and to allow the tracking value to be
... specified, a typographic term that refers to the spacing between characters. There's a
... related issue called kerning. It turns out that one of the Japanese deployed specifications has
... an attribute to turn on and off kerning. Japanese doesn't usually use kerning, but it may be
... useful for embedded latin scripts. To support that we need to have a property font-kerning
... as specified in the CSS3 font module - I'll add an issue for that.

<scribe> ACTION: glenn Raise an issue for font-kerning in TTML2 [recorded in http://www.w3.org/2014/12/18-tt-minutes.html#action01]

<trackbot> Created ACTION-362 - Raise an issue for font-kerning in ttml2 [on Glenn Adams - due 2014-12-25].

close issue-236

<trackbot> Closed issue-236.

issue-237?

<trackbot> issue-237 -- Inline space -- pending review

<trackbot> http://www.w3.org/AudioVideo/TT/tracker/issues/237

glenn: This allows the author to define a space in the inline progression direction.

close issue-237

<trackbot> Closed issue-237.

nigel: Last week we could not close a number of issues related to the ttm:item solution.
... There has been some discussion - I'd like us to get to sufficient consensus for FPWD, even if
... later we modify it based on further discussion, e.g. prior to CR.

glenn: There were 2 issues raised - one related to validation of the content of the item element,
... noting that we don't have a constraint that all spec items can be validated by XML Schema.
... Though I guess that a constraint based on the content type of ttm:item based on the name
... and a lookup may be validatable using Schema 1.1.

nigel: Really?

glenn: It would have to have name -> type dereferencing logic. I haven't tried it.
... The other issue is regarding the value definition of names. Nigel proposed that we should
... find a way to [scribe misses audio]. The most appropriate mechanism I think is to use
... the xml namespace mechanism, to allow a prefix to qualify the name, where the prefix
... can be resolved to a URI that doesn't have to be reused every time (i.e. it's inherited). So I
... propose to change the syntax of name to NCName or an unqualified name - if an unqualified
... name then it would be implied to be within the TTML namespace definition, e.g. the "TT
... metadata items namespace". For people who want to define an alternate namespace for a
... defining authority or classification scheme, then e.g. "ebuttm:" could be used and then
... somewhere in the document associate that namespace with the prefix. Then the name value
... in ttm:item would be name="ebuttm:whatever". That's my suggested way to handle name
... extension mechanisms and to have a shorthand for the prefix mechanism.

Andreas: In general I think that's a reasonable concept proposal, and will be worth reviewing.
... For me, this is a complete concept to integrate metadata in TTML without dependent on the
... TTML publication process. I can see the reasoning to find a way to handle metadata independently
... of the rest of the specification. I don't want to block the FPWD process though I think it
... does need more review. I think this should be put into the WD and should be reviewed by
... e.g. EBU to see how it matches with the current mechanism. Regarding validation, it would
... be good to have a position about it, how important current validation mechanisms are for
... standardisations of new items. From a practical view some validation mechanisms are more
... used than others. I'd welcome other views on this as well as mine - are there other users
... whose workflow depends on validation?

glenn: Thinking of the practical consequences of e.g. using XSD formally, and then we say that
... we're going to interpret every TTML constraint that can be expressed in XSD then that would
... be quite a task even for TTML1. For example the rules on the timing expression and a variety
... of other expressions are just xs:string because writing out a regular expression or schema
... syntax for some of the more complicated attributes may be extremely difficult - e.g.
... length and time expressions. Regardless of RNC or XSD we didn't go as far as possible. We
... ended up incorporating some of those - e.g. textDecoration has all possible combinations
... enumerated in the schema, but in other cases it's difficult. The practical consequences of
... pushing XML schema more. And we didn't have TTV for example available at that time. There's
... a place for validators and verifiers that can go further than schema validators.

pal: For metadata that is purely descriptive, why even create an equivalent in TTML2 and not
... just encourage metadata to be added in foreign namespaces?

glenn: We don't say anything about that - we imply that it's a reasonable approach. The reason
... I chose to try to pull in the existing metadata that I could find between SMPTE-TT and EBU-TT
... was because I think a lot of it has general utility to TTML and having it defined in one place
... is a good thing to do. If we had gone down that path in TTML1, when we reviewed some
... specific metadata items from Dublin Core and MPEG-7 we decided we weren't ready to
... define them. Since then we've deployed, different groups have proposed specific metadata
... items and I think now's the time to bring them into TTML unless there's a good reason not to.
... That allows them to be commonly used.

pal: I appreciate the goal. I see a danger: e.g. CEA608CaptionService. The definition of that item
... is substantively different than RP2052-10 that's supposed to be equivalent. How can we
... avoid confusion between those two parameters that look like they're supposed to be the same?

atai: I see the same danger for EBU-TT vocabulary.

glenn: I want to know what the difference is because I tried to stay faithful to the existing
... definitions where I put the prefix cea on them - I added that prefix because I recognised
... that the way it was defined in the RP you referred to was quite specific to CEA608 and I
... didn't think it was appropriate to make it more generic. I haven't yet refined the value type
... definition to be more refined than e.g. xs:string. You can view all of the xs:string ones as a
... "tbd" to plug in the more specific definition.

pal: Okay, I still see the danger of having two metadata definitions that could both be present
... and have different values.

glenn: I'm expecting a translation process. It's possible just to reference RP2052 directly,
... but then we'd define very few values or have to normatively reference downstream specs,
... which I've commented on before. We could argue that RP2052 is not downstream from TTML2
... but from TTML1 so technically that might not be a circular reference, but if it ever changed
... to refer to TTML2 then I'd still have that concern.

pal: We could resolve that by communicating with SMPTE who could act on that. There's
... another possibility, which is simply to do nothing, and leave those metadata items in SMPTE
... and non-normatively point the reader at them.

glenn: Yes, that's an option, or we could define something that points at the SMPTE document
... to get the value definition.
... (with a normative reference)
... Other than the circular reference problem I don't see any problem with that.
... I also see a value in enumerating the values here, as a single point of reference.

atai: I agree with pal's argument - I would favour a solution where the TTWG, SMPTE and EBU
... find a way to coordinate their metadata efforts to create an interoperable solution without
... conflicting definitions or the possibility of confusing duplications.
... On validation mechanisms, not everything can be validated in Schema 1.0 but specific
... constraints like the existing of metadata elements could be validated with Schema 1.0 and
... that would not be the case with the proposed solution.

mike: It's important that even if there's a general way of adding metadata we are clear how to
... replicate what's in 2052-10. There may be a transitory period where there could be some
... confusion, until the responding organisations update their specifications to reference the TTML2 work as needed..

glenn: There's a different point about bringing things into TTML and then revising external
... specs. The reason I brought these into TTML2 was because nigel, mike, andreas and pal had
... followed a bunch of issues saying we should support these metadata. Now it sounds like you're
... saying you don't want a new mechanism.

mike: For SMPTE and DECE the intent was to sweep the metadata into TTML2 and then reference
... it, removing the metadata from the SMPTE and DECE. That's the long term plan. The requests
... from SMPTE and DECE to add these items was for that purpose. Provided the semantics are
... the same or at least subsettable then that follows the intent.

atai: I'm not sure about the EBU requests for adding metadata vocabulary - I think we have to
... go back and look at that. I thought the request was to add the metadata in the ebu namespace.
... EBU needs to consider this further. Also there have been some comments from John Birch
... on the list. For me, I would abstain and we can proceed with this and then review by other groups.

glenn: It seems as though all the issues can be surmounted by adding more spec text.
... If there are semantic discrepancies then they can be fixed. I'm not hearing any strong objection
... to moving forward with that in the FPWD. I propose closing these issues en masse and then
... reviewing post FPWD.

nigel: I was going to make that proposal myself.

pal: I have a problem - the FPWD doesn't need to be perfect, but the definitions should at least
... be a superset of the SMPTE ones - that needs to be fixed before the FPWD. That doesn't
... sound like a lot of work.

glenn: Please tell me what's broken specifically.

pal: RP2052-10 section 5.6 has a specific definition of a caption service - that doesn't exist
... in TTML2.

glenn: There's an Editorial note to further specify the details of the named items.

nigel: There's clearly some editorial work to be done as described by Glenn earlier, as well
... as the editorial points on each metadata items.

glenn: This is a blocker as raised by pal. If we insist on this then it will push the dates back
... by a month or so. I wanted to get all the main Change Proposals marked as implemented.
... That doesn't prevent us from dealing with all the issues.

nigel: I was referring to the technical changes - I think that the editorial notes are okay for a FPWD
... as long as it's clear where work is needed.

glenn: I agree with that - or we could remove all the items, but I think that would be a disservice.

nigel: Given the state of this discussion and the outstanding technical edits I can't close these
... issues at this time, so I'll have to hold them open until the new year.
... Apologies for not getting to the other business. Wishing everyone a good couple of week's
... break if you're getting one, and see you in January. Thanks for all your work this year.

glenn: Mele Kalikimaka! - Merry Christmas in Hawaiian!

nigel: yes, Happy Christmas too. [adjourns the meeting]
... Oh, forgot to add, since we closed Issues 236 and 237 we can close CP20.