See also: IRC log
<scribe> scribe: nigel
Nigel: For today, we have 2
hours. I don't think there's anything to discuss re TPAC,
so
... I think we'll be focusing on TTML1 progress, issues, pull
requests etc. Do we need time
... on IMSC?
Pierre: Yes, should be very quick but there's a blocker there.
Nigel: I think the HDR in PNG
note has had its time, and I haven't had any responses,
so
... let's discuss publishing that as a Note.
... Any other business or particular points to raise?
Pierre: Dae didn't you propose resolutions on a bunch of issues?
Dae: Yes but they're waiting for feedback from Richard Ishida.
Nigel: I don't have any other business, but just noting today is Global Accessibility Awareness Day.
Pierre: There's a pull request:
Added explicit spaces in tts:aspectRatio and tts:activeArea definitions (issue #221)
Attribute syntax definition: missing spaces
Pierre: In TTML1 there's prose
that has not been imported into IMSC1 about spaces being
... required between non-terminal tokens. It was never in IMSC1
even though it followed
... the conventions. So here I propose a pull request that
explicitly adds a single space
... between the tokens in ittp:activeArea and ittp:activeArea.
However Glenn is not happy
... with that.
Glenn: I think many readers would
assume that the TTML1 rules already apply even though
... it is not explicitly stated. By introducing a space there
we are going against how
... attribute syntax is specified in TTML1 and TTML2 since it
does not do that except to
... call it out when those rules do not apply. Also the way it
is quoted as a double quoted
... white space is ambiguous because we cannot tell which
character it is. Also it does not
... work in all contexts even with XML since it is possible to
escape white space and it may
... not be normalised. So the way the comment has been resolved
is inconsistent and it is
... much easier if you simply refer to the TTML1 rule in the
convention section or something
... like that.
Pierre: Nigel, what's your take on that?
Nigel: This is new to me, however
for the reason that implementers should not need to
... create a special attribute parser just for these two
attributes, I would agree with Glenn.
Pierre: Okay I'll do that. I'll
need to copy the text in otherwise I would have to reference
the
... two places in TTML1 where this is defined, one for
parameter attributes and the other for
... style attributes? Oh actually both are parameters. Thanks,
I'll go ahead.
Nigel: There are two other pull requests for IMSC - do we need to discuss those?
Pierre: We can do that offline,
no need to discuss today.
... Also FYI there are two issues Nigel that need your review,
and everyone in the Group's,
... for CR exit criteria.
Glenn: Question: Pierre, how did
you resolve that issue about the white space and the
... fontFamily expression?
Pierre: That's still under
discussion, there's no pull request on that. It's on the IMSC
test
... suite not on IMSC1.
Glenn: Your point was based on a presumption about TTML1 that may or may not be true.
FontFamily009.ttml has a spurious space
Pierre: That has an impact on TTML1 and TTML2.
Glenn: The issue is whether or
not space is permitted around terms in a comma separated
... list, and even more generally whether space is permitted
around other delimiters, of which
... we have a variety. That generic language we were talking
about for an implied whitespace
... between two non-terminals: if you have another kind of
delimiter there then that language
... does not read on the question of if it is permissible to
have white space between the delimiter
... and the term that follows. In this particular issue someone
had read it as indicating that
... white space is not allowed in that context but if you look
at the original CSS2 grammar
... specification from 1998 and this spec you'll find examples
where white space appears
... around terms in a delimiter separated list, and the same in
XSL-FO. We did not call this
... out in TTML but there's a good case for saying that white
space is permitted.
... This will affect all the TTML-based languages.
Pierre: I've arrived at a different implementation from Glenn here.
Nigel: Do we have example data
points here? I would expect to see "comma space font
name"
... in real world examples, and I'm sure I've seen that.
Glenn: I think the original DFXP implementations and TTV does permit it.
Pierre: It would be helpful and
relevant to see example content here. RGBA values with
... white spaces are more complex to process.
Glenn: I don't think the spec argues this clearly either way.
Pierre: There are no IMSC tests except for the fontFamily one that demonstrate this.
Glenn: There are some older tests for DFXP that I can go back and look at.
Pierre: At least 2 people in the
world read it in a particular way, so it would be a
... disservice to go back and change TTML1 to make
implementations incompatible unless
... we can show that many documents exist that have that
issue.
Nigel: There's an example at
http://bbc.github.io/subtitle-guidelines/#tts-fontFamily
that shows comma separated lists.
... I just provide this in the spirit of a single data point.
Look at the second row in the
... document requirements section.
Pierre: We may extend this to
fonts but I'm not comfortable extending it across the
board.
... This is about not making existing implementations stop
working just by a stroke of the
... pen on the spec.
Nigel: I would go back to my
previous point that it should be possible to have a
single
... attribute value tokeniser algorithm rather than being
required to use different ones.
Glenn: I don't think we can
resolve this but I'd point out that the new conventions that
just
... went into TTML2 explicitly allow for white space around
commas and semicolons where
... they are delimited. That would extend to all argument lists
that use commas for example.
Pierre: Unless there's a really
compelling reason to do that I'm not sure why we'd need
to
... broaden the scope there and increase the number of test
cases. Also we would need to
... be careful about terminal and non-terminal values, what
about time expressions?
Glenn: No, that's explicitly dealt with; white space is prohibited from those time expressions.
Pierre: Not in TTML1. Unless
there's a really good reason, regardless of CSS, we
should
... remove flexibility rather than increase it.
Glenn: For implementers that
assume that whitespace is permitted, and authors who
... assume it is permitted then your position might make their
assumption invalid.
Pierre: Only for authors. If we
are going to make that change, we need to know if authors
... have been using that capability.
Glenn: So we need more input on this.
Nigel: I agree we cannot resolve
it today.
... Specifically we would want to know if there are
implementations that do not work properly
... in the case that spaces exist, and if there are documents
that include additional spaces.
Nigel: Looking at the milestones,
we have 36 open, 47 closed on the Editor's list for WR,
... which is 56% complete.
... And we have 5 open 0 closed on the Group's WR action
required list.
Glenn: Approximately 22 have been
closed in the last week. If I can keep up this rate then
... we are going to be pretty close to getting to 100% by the
end of this month, so that is
... my hope.
... I noticed that Glenn has offered to work on #20 and has
offered a pull request,
... thank you for that. I'm working on an update to that pull
request today. Can I assign
... that to you Pierre?
How to specify aspect ratio to understand positioning that may apply for display or video
Glenn: This Pull Request will not be merged instantly.
Pierre: I don't think 3 days would be adequate for this one.
Glenn: [assigns issue to Pierre and Glenn] We're both working on this right now.
Nigel: What's next on everyone's slate for issues to pick up?
Glenn: I'm actively working on a
number that I am working on. I cleared a number of the
... assignments to myself so that others could feel free to
work on them. Since then I have
... picked up a few of them.
... In one case I submitted a pull request that was assigned to
you Nigel.
Nigel: I noticed that!
Glenn: Dae, I have a follow-on
request on one of the Ruby examples that I'd like to ask
you
... to fine tune so I will coordinate with you on the side on
that.
Add examples of new features commonly used with Japanese content (2 of 2)
Glenn: Also I'd like to point out
that there are number of places in the spec where
examples
... are needed, so Dae if you have a chance and some time we
would certainly welcome
... more submissions on those.
Dae: I'll give the document another pass-through and see what I can pick up.
Nigel: There's been activity on #128.
The uniqueness of xml:id needs to be broken for some uses of condition
Glenn: I'll try to add some data
points to that today.
... I'd appreciate it if you could pick up #224 and #156
Nigel.
Nigel: Okay I'll look at those and see if I can take them on.
Glenn: Also #135 clarify timing concepts in relation to SMIL.
Nigel: We did do some of that
already, and there's a note on the issue for Pierre to
add
... something about container diagrams.
Glenn: At one point there was an ask to more formally define the times than in SMIL.
Nigel: We did the work already in
the wiki, perhaps it just needs to be validated and
formatted
... into a pull request.
Glenn: My worry is it will be difficult to validate.
Nigel: In that case I would
propose a WG Note explaining how we interpret SMIL in the
context
... of TTML.
Glenn: I think that's a good
idea, and could be added to the spec later. We need to do it
in
... a quasi informative manner that can be validated rather
than making it normative spec
... text that may unknowingly break things.
... I think all of us have done some implementation work on
this and came up with different
... implementations...
Nigel: That's a worrying sign.
Glenn: The amount of test content in relation to timing is pretty minimal.
Nigel: I have a lot of cases that
can be generated from the EBU-TT Live Interoperability
Toolkit's
... BDD tests. That can be done for the TTML2 test suite though
they are subject to some
... starting constraints such as parallel only time container
semantics, and dur only permitted
... on body, but they would generate a large number of test
documents and expected
... processing outcomes.
Glenn: I will also need your help
on #96 the extension to add date support to time
expressions.
... I need to read the comments that came in from Nigel and
Addison. I'm reticent to putting
... back in time zone support but I am open to persuasion.
Nigel: Okay. I will review that
and see if I can add a further comment.
... Any further points on issues?
Dae: I will be taking a look, and
need to prompt Richard Ishida on some of the open issues.
... How long should we give the commenter to respond, if we're
hitting the end of June?
... Would 7 days be reasonable?
Nigel: From the Process
perspective we can take a view on each issue and we only
have
... to dispose of all the wide review comments for transition
to CR, so we could reassess
... the state say in the middle of June.
Pierre: I have a slightly
different view - if we take weeks to respond to a comment and
then
... give a short 7 day deadline that might seem funny.
<inserted> Dae: This is specifically about issues where the commenter and I are actively going back and forth and one side stops replying
No mention of isolation for directional controls
Nigel: This is assigned to you Pierre, and the last comment is from Richard.
Pierre: This is a new feature of
CSS with low levels of support; unless what we have today
... really does not work my view is to defer the feature to
TTML.next.
Glenn: This has already entered the Unicode bidi algorithm.
Pierre: I'm not the best person to comment on this.
Glenn: Also Unicode added
explicit control characters with the isolate semantics, not
in
... the bidi spec. The original reason for the bidi properties
in CSS were to provide a stylesheet
... mechanism that would express the same information as was in
the explicit bidi control
... characters that would allow authors to avoid using those
characters. Given the current
... state since it is permitted in the character content that's
a case where you could not
... represent it in style information in TTML right now whereas
in CSS you can. I was going
... to suggest that we accept this change and add the necessary
properties.
Pierre: I would not object to that.
Glenn: I do agree it would be work for implementations and testing.
Pierre: imsc.js would support it to the extent that browsers support it, which is spotty.
Glenn: The community will demand it so it will come along.
Pierre: Hopefully we'll have better support when we get to PR.
Glenn: I'm adding a comment to say that I generally support this.
Pierre: I can generate a pull request for it then.
Glenn: Sounds good to me.
Nigel: We need to have at least a placeholder for an example there too.
Pierre: People interested in aspect ratios etc should look at the pull request I drafted yesterday:
Initial pass at rationalizing aspect ratio and pixel dimensions
Glenn: As I mentioned before I am
in the process of creating a new commit for this branch
... which will generate some further discussion.
... This will knock off at least 2 or maybe 3 related issues
which I think I've already
... incorporated into #30 by reference at this point.
Nigel: Regarding mediaOffset Glenn and I had some email discussion.
Glenn: I need to read your
message carefully on that so it's pending. We are not
... synchronised in our understanding at this point. I don't
have anything specific that I
... need input on right now.
Specify date as well as hours, minutes and seconds in time expressions
Nigel: In this case I think we
need the option of a timezone because there are two
possible
... valid use cases for clockMode="local", one in which just
whatever local timezone applies
... is used, so the timezone is not specified, and the other is
that a specific timezone is
... intended and needs to be specified.
Pierre: How would you encode the timezone?
Glenn: Using the same syntax as in the source for this, which is SMIL, and the timezone part is based on ISO 8601.
group: Discussion of the merits
and ambiguity/clarity of including a timezone component
... in local times - Nigel argues in favour on the basis of
operational utility, understanding
... that a local time with a timezone offset is a different way
to express a UTC time;
... Glenn and Pierre argue that if a UTC time is intended then
a UTC time should be used
... and that local times always need to be treated locally.
Nigel: In that case if we are going to use local in this way we need to clarify it.
Glenn: That seems reasonable.
Nigel: I've added a comment on the issue.
Add ttm:mediaTimestamp (or equivalent) attribute
Glenn: I need to review your
input Nigel.
... I also need to understand the proposal to use set in
#128
The uniqueness of xml:id needs to be broken for some uses of condition
Pierre: I also need to consider
the impact on implementations there. The bigger question
... of when condition is evaluated was brought up, which is
somewhat orthogonal but
... something we need to consider at some point. It impacts
more than just this issue.
Nigel: I agree - this was one of my earliest questions regarding the condition mechanism.
Glenn: My current position is
that it needs to be resolved at presentation time. Also
there
... is a special case that can be made that some of the
semantics of a style element should
... not be ignored even though it is conditionally excluded.
Because of chained referential
... styling you might have two or three styles that refer back
to something that gets incorporated.
... If you conditionally exclude a style in the middle of a
chain does that break the chain
... or just exclude those style attributes defined by that
style element?
... This is a case where partial semantics need to be retained.
I've implemented that in TTXV.
Nigel: This points to a problem
possibly with the use of the term "ignore" in that it
does
... not mean that the element itself is ignored but that the
things it defines are ignored.
Glenn: Yes that is exactly the case.
Nigel: So in terms of action we need to change the term ignore or define it better.
Glenn: I agree that more clarification is needed.
Pierre: It has to be added to the style resolution process.
Glenn: I agree.
... I have some tests in the TTT test suite that test this.
action-495?
<trackbot> action-495 -- Thierry Michel to Update the ttwg homepage and publications pages for the new repos -- due 2017-04-20 -- OPEN
<trackbot> http://www.w3.org/AudioVideo/TT/tracker/actions/495
Nigel: I see that this is done.
close action-495
<trackbot> Closed action-495.
PROPOSAL: Publish the PQ HDR in PNG document as a Working Group Note
Nigel: I've just spotted that the ICC file claims that it has copyright with Adobe Systems.
Glenn: Apache licensing doesn't
allow use of an ICC file; we had to make other
arrangements.
... It was copyright by Adobe.
Pierre: So Adobe holds a copyright on every ICC file ever made?
Glenn: No, there's a standard one
that's widely used in the industry but Apache decided
... we could not include that in any Apache project.
Pierre: This ICC file is bespoke
and was made by hand.
... Thanks for highlighting this. It was not taken from any
existing tool.
Nigel: Presumably we need to update the copyright notice here to say W3C.
Pierre: I will follow up with whoever created it.
Nigel: I think we can resolve to publish that after making those changes.
Pierre: Okay, if I learn that we
cannot change the copyright notice on the ICC I will
notify
... the group.
RESOLUTION: After applying fixes to the outstanding issues, publish the PQ HDR in PNG document as a WG Note.
Nigel: Thanks everyone. [adjourns meeting]