Timed Text Working Group Teleconference

<scribe> scribe: nigel

This Meeting

Nigel: For today we have some updates on IMSC and some TTML issues. Any other business?

group: [no other business]

Nigel: Next week I will not be available, and nobody has offered to Chair. Any volunteers?

Andreas: I may also not be able to attend next week.

Nigel: If anyone does want to Chair next week's meeting please let me know by Tuesday
... at the latest and I will be able to assist with any queries etc.
... Otherwise the default position may be that we do not hold a meeting.

IMSC

Nigel: My update is that after publishing the WD for Wide Review Thierry sent the horizontal
... review and internal dependency review messages out to other W3 groups.

Thierry: I have sent the questionnaire to the group and that is holding back the request to
... TAG. That's the only one that is pending.

Nigel: Thank you. And I sent all the external liaisons. It turns out that DLNA is no longer
... operating, so I could not send to them. I got several responses back from people saying
... they had received it and would provide feedback.
... Thierry sent that questionnaire out on 24th March. Any queries on that now?

Thierry: I reused the wording from TTML. Probably what may be missing is stuff about the
... use of images in IMSC1.

Nigel: I see that the answer to 3.6 needs editing to remove audio and condition for example.

Thierry: Ok

Nigel: If you make those edits then we can approve the updated text by email. I think the
... remaining responses are uncontroversial.

Thierry: I will do that right now.

Nigel: We had one issue raised on IMSC this week that is of some interest:

https://github.com/w3c/imsc/issues/224

Nigel: This is about whitespace handling.

Glenn: The issue here is that the wording of XSL-FO makes it hard for me to determine if
... the intent is to fully translate all whitespace that is XML whitespace into 0x20 space characters.

Nigel: I had the same issue, I wasn't sure exactly what it was trying to say or if there is some
... order of precedence in applying the various rules.

Glenn: The problematic part is whether or not collapse semantics apply, and it seems like
... you can read it that for the first character in a sequence of whitespace, if it is not a space,
... then you might not collapse it. I do not think that is the intent and I do not know of any
... implementation that does not translate tab in that context into space, so I think implementations
... should collapse it.

Nigel: I think my colleague Chris (who raised the issue) has an implementation that seems
... to have this behaviour of retaining the first tab, and seems to obey the letter of the
... specification rather than the intent.

Glenn: I think the Antenna House and FOP implementations do it differently, so the question
... is I guess what we really want to happen. If we want that tab to be normalised to space
... then we need to fix up our spec. My suggestion is to add language to TTML1 and TTML2
... to have that effect. Even if we do not translate it into space there is certainly no suggestion
... that it be translated into an alignment or some spaces.

Nigel: Even if no alignment and no spaces a tab would be expected to have some inline width.

Glenn: Usually a tab character is not in a font and is mapped to zero width. It has no
... inherent width semantics.
... Probably any fonts that do have a tab character map it to zero width or to the space glyph.

Andreas: I have a question regarding Chris's original question. He asked on the basis of
... Example 7 of IMSC 1.0.1?

Nigel: Yes.

Andreas: So actually we did not provide the source for the example?

Nigel: Yes, we do.
... The way the spec is structured the example is an XML document that is imported by the
... HTML and then formatted. So the tabs are really there.

Glenn: We should fix the example.

https://www.w3.org/TR/ttml-imsc1.0.1/#itts-fillLineGap

Nigel: Yes I agree and Pierre has indicated he agrees with that, on the issue.

Glenn: I agree with Andreas however, that said, I think we should remove the tabs.
... There's another property called whitespace-treatment ...

Nigel: There are 4 properties defined by TTML1:

https://www.w3.org/TR/ttml1/#content-attribute-space

Glenn: If you look at §7.16.8 in XSL 1.1 on whitespace treatment:

<glenn> https://www.w3.org/TR/xsl/#white-space-treatment

Nigel: Since we use suppress-at-line-break=auto that means only U+0020 is suppressed,
... and everything else is retained.

Glenn: Correct.
... Under §1.2.1 under paging and scrolling it talks about splitting things like whitespace
... into separate properties for greater control. The intent was to derive those properties
... from CSS2, in this case the CSS2 property called "whitespace". So to follow the implementation
... path of other implementations we probably want to go and look at CSS2 whitespace.

<glenn> https://www.w3.org/TR/REC-CSS2/cover.html#minitoc

Glenn: The initial value of whitespace is "normal" which directs user agents to
... "collapse sequences of whitespace, and break lines as necessary to fill line boxes"
... and does not make a distinction between different kinds of whitespace.

Nigel: If you follow the link it defines whitespace as:

-> "Only the characters "space" (Unicode code 32), "tab" (9), "line feed" (10), "carriage return" (13), and "form feed" (12) can occur in whitespace. "

Glenn: Exactly. These are the same as XML whitespace. It looks like there's a bug in XSL-FO here.
... I wonder if this was updated in a later version of XSL-FO.
... Perhaps Steve Zilles would be a good person to ask about this.

Andreas: I think we could also contact Tony Graham who is at Antenna House.

Glenn: We should certainly contact Steve as he is still active in W3C whereas Tony is not.

Nigel: Which should we do first, contact Steve or create the issues on TTML?

Glenn: I will create the issues and link them to the IMSC issue from Chris and articulate
... the problem as I see it and then we can use that as reference material in an email to
... Steve and Tony.
... I know both Steve and Tony so I can ask them. Since I chair the Apache group on FOP
... I have a link there too.

Nigel: Okay, that's a plan, thank you.
... I will add a note to the issue.

https://github.com/w3c/imsc/issues/224#issuecomment-290434056

TTML

https://github.com/w3c/ttml2/issues/233

Nigel: This raised some eyebrows, since there was already a pull request that Pierre had
... opened on request.

Glenn: I always edit pull requests before applying them to add my imprimatur and make
... sure they are consistent.

<tm> using PR is not 10 yeras of process ...

Nigel: From a process point of view, we had https://github.com/w3c/ttml2/pull/283 which
... the Group requested from Pierre to resolve the issue he raised, which was closed without
... any communication and it was not obvious why. The branch for the pull request could
... easily have been modified in place rather than closing and opening a new one, which
... would have been less work and would have made the provenance clearer.

Glenn: One thing here was that the branch was not properly named and had things in
... which were not done in the right way. There was no intention to raise tensions.

Andreas: This is an interesting discussion - things like GitHub have changed the way that
... groups can collaborate on specification texts. I think we appreciate that the Editor here
... has a different view maybe. We as a Group should decide on what the process is that
... we use, and also recall that proposals and changes can be made onto the pull requests.

Nigel: By the way you can rename a branch if that is the issue.

Glenn: It is not so straightforward because the pull request doesn't get modified. It is okay
... to do it locally before pushing to the remote, but when pushing to the remote things
... get locked in.
... The only substantive point on this particular issue is the name. I've stated my opinion
... on that. I don't think adding length to the name has any improvement. That's my opinion.

Nigel: The name needs to reflect the scope of application, which is much narrower than it
... is specified at the moment, and appears to apply to all HDR content rather than
... just those HDR schemes that require it.
... This also needs to be reflected in the specification text, which likewise needs
... to be narrowed to schemes such as PQ HDR.

Glenn: Looking at the original proposal, that referred to images, which I took out.
... I took "image" to refer to the content on which it is composited rather than a referenced
... PNG or JPEG image in an <image> element. Is that not right?

Nigel: No that's correct I think it is about the video images, for example.

Glenn: Pierre used the term "target image" and I changed it to "target medium" - I found
... ways to eliminate the use of "image".
... The other thing is that the example has the unexplained value 2.4 so I added "where the
... exponent of 2.4 is the gamma correction applied by the EOTF". I may have been wrong
... in that but that was my interpretation.
... Just below the syntax box there is an equation that uses the term "normalized" which
... I assume meant application of the gamma of 2.4.

Nigel: That's the root of the problem - there is another scheme for HDR that does not use
... the luminance gain approach at all but uses a different approach varying the gamma.

<glenn> https://en.wikipedia.org/wiki/Gamma_correction

Andreas: I have to drop off now. Thank you.

Nigel: By specifying tts:luminance and using this kind of wording it looks as though we are
... promoting one particular scheme for HDR, which is not what we should be doing at all.

Glenn: I was not aware of that context.

<tm> OK bye

Thierry: [drops off call]

Glenn: I did not make any other changes here.
... Actually I did fix another change which was an error - he had an integer type but in the
... examples used non-integer numbers so I changed to a non-negative real number.
... I also added continuous animation support. Any numerical value I make continuous.

Nigel: By the way the other colour space that is often used is BT.2020 - we currently only
... refer to BT2100 in Appendix P. So there needs to be some work here to make this
... more neutral.

Glenn: Please note that people file new issues on the text rather than piling comments on
... to the original issue. We can always point back to the original issue.

Nigel: Can I just check - are there other branches with work in progress on them that
... are not yet complete and might be worth review?

Glenn: I have about 10, particularly Audio Description. I'm hoping we do not slide too far
... beyond the end of the month. The other one that is higher up on my list is the one that
... came up last week regarding Andreas's pull request on TTML1.

Nigel: I see that just Glenn and I remain on the call so we should probably close!

Glenn: I will take a look at Andreas's pull request and may need to make some changes to it.

Nigel: If you can edit the branch that is the source of the pull request that will be better.
... Even if you commit directly to that branch.

Glenn: Okay I will give that a go.

Nigel: Ok thanks, see you in a couple of weeks. [adjourns meeting]