TTWG minutes 11/21/08

TTWG minutes 11/21/2008.  Note action items below.

Attending:
Sean Hayes (SH; chair)
Geoff Freed (GF; scribe)
Glenn Adams (GA)
Philippe Le Hegaret (PH)
Andrew Kirkpatrick (AK)

Regrets:
David Kirby (DK, chair)
Mike Dolan (MD)

======

SH:  Action items-- there are four actions in tracker.  Can't add any  
new actions yet.  Would like to get new actions into tracker so we  
can track them.

PH:  Use this for issues and action tracking.  We have two open  
issues now. (Pastes into IRC.)

SH:  Can we get all other issues?  They've been circulated in past  
agendas.

PH:  I can put them in the system.

SH:  Open actions:  SH sent work statement on SMPTE to the list.   
Close this action.  Close #16 as well.  PH must look into default  
values for styling.

PH:  Not done yet.  Will do.  That issue is to investigate what  
defaults apply to style properties.  tts:color default=transparent,  
for example.  CSS is not implementation-defined.

SH:  Will send other actions to PH.

ACTION:  SH to send other action items to PH.

SH:  I've been out.  Can someone summarize where we are now?

PH:  Good progress so far.  Glenn's test suite is too complex for any  
player but his.  We're going to re-write test suite to be simpler for  
other players (CCforFlash/ccPlayer or Adobe's player for example).   
Created testing framework, which was sent to the list.  Now we need  
to divide the creation of test files among group members.

GF:  ccPlayer works fine, not JW Player.

PH:  Must run JW Player from a Web server.  Will check with author to  
see if that's a problem that can be fixed.

SH:  May have that same problem with SilverLight Player, too (won't  
run locally).

PH:  Workaround will be to run the testsuite from a Web server.  Not  
a problem to do this.

AK:  Can change an HTML parameter to make it work-- take a look?

PH:  Will investigate.

PH:  Send examples of SilverLight and caption file and I will add it  
to the test suite.  Also send Adobe example.

PH:  Zip file doesn't contain a SWF, but GF indicates that we can  
include ccPlayer.SWF.

GF:  Yes.

SH:  So where are we in our schedule?

PH:  Must complete the test suite (two versions, presentation and  
transformation).  This will be a subset of the full spec.   
Transformation set:  everything still needs to be done.  Do we want  
to do SRT, SCC, 708, other formats?  And the DFXP spec needs to be  
updated.  We need to have a default region in the spec. Can't really  
specify what the default region looks like, however.

SH:  I was supposed to find commonalities in styles between UK and US  
captioning.

PH:  We should leave it to the user agent to define a default style,  
though.

SH:  Fine.

PH:  Would be most appropriate.  Leave it to implementation.  Player  
should just show the text, wherever the player defines a default region.

SH:  So we should say there must be a default region, but not define  
its style.

GF:  Due date?

PH:  Last call is January 2009.  Test suite should be nearly complete  
by this time.

SH:  We should integrate the open spec issues into the issues  
database, then use the survey tool to resolve these quickly.

PH:  Discuss xml:id now?

PH:  NCAM and Adobe use id and not xml:id.  Future releases of  
ccPlayer/CCforFlash will use xml:id and id.  What about Adobe?

AK:  Not sure yet.  Will look into this.

ACTION:  AK will investigate the use of xml:id.

GA:  XFSI had supported id earlier, switched to xml:id.  Have  
reservations about using xml:id in the first place.  The support of  
both might be the best compromise.

PH:  Should be fine to support both.

SH:  No objection.

PH:  DFXP must say that authors must use one or the other.

SH:  Could use xml:id and id on different elements, or should we  
specify one or the other within one document?

PH:  Use both but must have the same value?

SH:  Not sure.

SH:  Problems with xml:lang?

PH:  CCforFlash supports xml:lang.  Not sure of others.

SH:  May not be an issue.

PH:  Need to focus on spec issues and test suite to make last call in  
January.

SH:  Send any spec issues to PH to get them into the tracker.  In two  
weeks we'll review and create a survey with those issues and a  
proposed resolution for each.  Knock them off and get them into the  
spec.

PH:  Running into some bugs in CCforFlash.  Will send GF notes.

SH:  More questions on testing?

AK:  Re default region-- are we saying that test files must indicate  
a region?  Our player assumes a region defined by the player.

PH:  No region should be defined in tests, as they won't work in any  
player but Glenn's.

AK:  Okay, but should also write tests to test region and multiple  
regions.

PH:  Right.

PH:  Another issue:  xml:space attribute.  Adobe ignores it, NCAM  
ignores it.  Spec default value is preserve (white space).  If you  
write text with <br /> and hit return, you will have two carriage  
returns between rows of text.  Should only be one.

GA:  Is a bug in the implementation that should be changed.

SH:  So the implementations are failing that test?

PH:  Should add xml:space to our list of issues.

SH:  Don't think xml:space is a bug in the spec, but seems that  
players are implementing it incorrectly.

PH:  Right-- they're not following the spec default value.

SH:  Probably pertinent to add this to a category of bugs in players,  
but should not influence the way that we write the text.

PH:  Correct.

GF:  Should we assign the writing now?

GF:  Volunteers for style .

SH:  Volunteers for timing.

SH:  Parameters section is not big but is not clearly written.   
Should we volunteer DK?

GF:  Yes.  DK takes parameters

AK:  Will look for one to take.

GF:  How many style tests will need to be written?

PH:  We have 25 properties; tts:color is finished.

AK:  Will split style tests with GF.

PH:  Should first concentrate on what NCAM and Adobe support.

SH:  Glenn, will do layout or metadata?  SH will add animation to  
timing.

GA:  Will take layout and animation.

SH:  Metadata may be straightforward.  Will ask DK to take this, too.

PH:  Can help with styling section.

SH:  Metadata:  agentname and actor may be complex.

ACTION:  Writing assignments as follows:

GF:  style
AK:  style
PH:  style
GA:  layout and animation
SH:  timing
DK:  parameters and metadata


SH:  Are we going to try to write tests for synchronic region/flow  
constructions?  No implementations yet.

GF:  You mean roll-up:

SH:  Yes.

GA:  That is different than dynamicText/Flow.  Is it in the layout  
section?

SH:  Yes (9.3).

GA:  That has nothing to do with dynamicFlow stuff.  It relates to  
specifying semantics in the context of timing and styling.

SH:  Then 9.3 can be covered in timing and layout tests.  Then that  
leaves dynamicFlow.  Are we going to create tests for that?

GA:  Only implementation now is from John Birch.

SH:  Currently dynamicFlow is under style.

GF:  Discuss dynamicFlow in the near future-- two weeks?  Three weeks?

SH:  Will add to agenda; perhaps extend the next telecon to cover  
both the survey and dynamicFlow.

PH:  Who will write tests for section 7?

SH:  Am in the process of writing a validator, so I could take that.

AK:  I have some files that show (...).

PH:  Had the idea to write in the caption itself to demonstrate what  
you should see on the screen if the test is working properly.  (See  
current examples in test suite.)  By reading the captions you should  
be able to tell if the test is successful or not.

SH:  Good for style but not for other areas.

PH:  Can use a counter in the captions for timing, for example (see  
test-suite examples).

SH:  Is largely structural, no?

AK:  There are timing issues around <body>, however.

SH:  Section 7 talks about styling elements in <head>, but attributes  
are discussed elsewhere.

PH:  If you nest spans, then the second span is ignored (NCAM  
player).  Might be a bug on NCAM's part.

GF:  Will check.

AK:  Spec says no nested spans.

PH:  But it does allow them.

SH:  Prose says you can't have nested span, but the diagram shows a  
nested span.  It's an inconsistency in the spec.  See 7.1.6, third  
paragraph.

AK:  It says explicitly that you can't have nested spans.

GA:  In an earlier version of the spec we said no nested spans, but  
then changed our minds.

GF:  Update to allow nested spans?

GA:  I think we should, from a language-specification perspective.

SH:  Have the prose reflect this.

ACTION:  Update spec to permit nested spans.

AK:  Is a good thing for us to have, but does anyone currently  
support it apart from GA's implementation?

SH:  Silverlight would.

AK:  Not sure of the likelihood of getting this incorporated by January.

SH:  Must get the spec right, but this may not be required of a  
presentation engine at that time.

PH:  We're going to write tests but some will not work at this time.

GA:  Should decide what features are mandatory in presentation  
profile, and this will dictate which tests in the suite are  
appropriate for that profile test.

SH:  So everybody get writing!

SH:  SMPTE captioning effort:  packaging group had a kickoff meeting  
two weeks ago; captioning meeting will start on december 4 in NYC.   
See agenda for link to sign up.  Must be a SMPTE member.

AK:  Did everyone see the article on captioning on A List Apart?   
Written by Joe Clark.

GA:  Send a link to the reflector.

AK:  Main argument is that we should just have open captioning.

AK:  Is somewhat off base on a number of points.

PH:  Another interesting problem:  Streaming DFXP is not possible  
right now.   If we make a presentation profile, the constraints to  
streaming may go away.

SH:  Can discuss on a later telecon, given the time.

SH:  No meeting next week; **next meeting is December 5**.  Main  
topic will be dynamicFlow.

Telecon ends.

Received on Friday, 21 November 2008 16:18:39 UTC