W3C home > Mailing lists > Public > public-tt@w3.org > October 2004

Minutes of TT WG Meeting on September 14-16, 2004, Amsterdam (CWI)

From: Thierry Michel <tmichel@w3.org>
Date: Tue, 19 Oct 2004 09:51:46 +0200
To: <public-tt@w3.org>
Message-Id: <20041019075146.0EBA04F48B@homer.w3.org>


Minutes of TT WG Meeting on September 14-16, 2004, Amsterdam (CWI)

Attendees

  Glenn Adams (XFSI, Chair, Scribe) [GA]
  Dick Bulterman (CWI) [DB], 1st day only
  Kaes Blom (CWI) [KB], alternate for CWI
  Sean Hayes (MSFT) [SH]
  Dave Kirby (BBC) [DK]
  Mike Evans (BBC), alternate for BBC, 1st day only
  Thierry Michel (W3C) [TM]

Observers
  Frans De Jong (EBU) [FJ], guest of chair, 1st day only
  John Birch (Screen Subtitling) [JB], guest of chair, 1st day only

Regrets

  Mike Dolan (Invited Expert)
  Geoff Freed (WGBH/NCAM) [GF]
  Erik Hodge (REAL) [EH]
  Dave Singer (Apple) [DS]

************************************************************************
Agenda
************************************************************************

Day 1 (Tuesday, September 14, 2004)

  09:00 - 10:30 Agenda Planning
		Introduction
                Review and Acceptance of Prior Minutes
                Report on EBU/AAF Subtitling
		WG Schedule Discussion
  10:30 - 11:00 Break
  11:00 - 12:30 Fill Mode Module Review [JB]
		DFXP WD Review - Organization
		DFXP WD Review - Content Vocabulary
  12:30 - 13:30 Lunch
  13:30 - 15:00	DFXP WD Review - Region Formatting
  15:30 - 16:00 Break
  16:00 - 17:00 DFXP WD Review - Region Formatting

Day 2 (Wednesday, September 15, 2004)

  09:00 - 10:30 DFXP WD Review - Region Formatting
  10:30 - 11:00 Break
  11:00 - 12:30 DFXP WD Review - Style
  12:30 - 13:30 Lunch
  13:30 - 15:00 DFXP WD Review - Style
  15:30 - 16:00 Break
  16:00 - 17:00 DFXP WD Review - Style

Day 3 (Thursday, September 16, 2004)

  09:00 - 10:30 DFXP WD Review - Timing
  10:30 - 11:00 Break
  11:00 - 12:30 DFXP WD Review - Metadata
  12:30 - 13:30 Lunch
  13:30 - 15:00 DFXP WD Review - Requirements Satisfaction
  15:30 - 16:00 Break
  16:00 - 17:30 DFXP WD Review - Style

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Day 1 (Tuesday, September 14, 2004)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
EBU/AAF Liaison
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[FJ] Presents liaison report on work of adhoc group organized by EBU,
AAF, and BBC to discuss common caption/subtitling authoring needs at
AAF/MXF levels.

[FJ] Sent liaison report to [TM] in June, but haven't seen response.

[GA] Apparently this report didn't make it to any of our reflectors,
so none of us were aware of it. Perhaps it didn't get to [TM]'s mail
box. Asks [FJ] to resend to public reflector.

[SH] Speaking as a member of AAF board, would like to see letter from
this group to EBU/AAF indicating expected schedule.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
WG Schedule Discussion
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Goal: 1st public WD of DFXP published by end of september.
Goal: Last Call of DFXP published by mid-november.
Goal: 1st public WD of AFXP published by end of year.

[TM] Reiterates need for high priority to produce public WDs ASAP,
otherwise any chance of recharter is slim.

[GA] Concurs.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Fill Mode Module Review [JB]
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[JB] Reviews fill mode module [1].

[1] http://lists.w3.org/Archives/Member/member-tt/2004Sep/0036.html

[GA] need "none" value for clear, re-flow, fill mode properties; need to
redefine "auto" as well;

[GA] will need to define an idealized model that permits precise
specification of content, flow, and timing, buffering, and presentation.

[GA] requests [JB] to make further progress on SVG examples that depict
operation of fill mode parameters.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
DFXP Review - Overview [2]
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[GA] Begins review of DFXP 2nd Draft at [2].

[2] http://lists.w3.org/Archives/Member/member-tt/2004Sep/0022.html

[SH] Wants to see good overview before technical material.

[GA] Suggests it be included in introduction.

[DB] Suggests putting definitions section into introduction.

[TM] Need to clearly demarcate informative and normative text/material.

[DB] Should have simple and not-so-simple examples in introduction.

[DB] Suggests adding "Overview and Motivation" informative text into WD.

[GA] Can include design notes, annotations, etc.

[GA] Need to add "Conventions" subsection under Definitions or Intro;
may aggregate conventions and notations.

[SH] Would like one line in abstract that indicates that DFXP will
be a proper subset of AFXP, which will have to be briefly introduced
here as well.

[DB] Suggests waiting until writing introduction before fine tuning
abstract.

Action: [GA] need to document conventions regarding normative definition
of element vocabulary and syntax used therein.

Action: [GA] need to characterize definitions of vocabulary as normative
while supplied schemas are informative.

Action: [GA] review recent proposed changes regarding xml:space from
XML Core Working Group with view of changing our vocabulary definitions
as necessary.

[SH] If we use the vocab definition conventions as our normative syntax
for defining vocabulary, then should be able to represent solely in
ASCII (e.g., use of bold font to indicate required attribute may be
problematic).

[SH] Why have multiple metadata elements as children?

[GA] Reviews earlier thinking:

<metadata ttm:actor="Pierce Brosnan"/>
<metadata name="ttm:actor" content="Pierce Brosnan"/>
<metadata>
  <actor xmlns="*foo*">
    <first>Pierce</first>
    <last>Brosnan</last>
  </actor>
</metadata>

[GA] Asks query about whether to take out <div/> or not; discussion
ensues. Some reasons offered to retain: allows logical organization into
units for authoring or other purposes; allows associating divisions
with regions rather than their children; allows separating region
style defaulting with parental division based style defaulting.

[DB] Good material to put into a "Overview and Motivation" subsection
for the div itself.

Resolution: retain <div/> in DFXP.

[GA] Makes an (as yet) unsuccessful case for not supporting nested
div.

[DK] Notes need for background color on span in order to supported
"robotic voice" style guide rules.

Tentative Resolution: support nested <div/> in DFXP.

Action: [GA] restore nested <div/> in content model.

Action: [GA] ensure that background color applies to content elements
as well as region.

Action: [GA] change language in vocab element defs that defines equiv
class of attrs in foreign namespace to not admit unqualified attrs.

[DK] need to be clear on what end time means.

Action: [GA] need to ensure that we document whether time intervals
are half-opened intervals or closed. We always use half-opened intervals
for [begin,end) = [begin,begin+dur).

Action: [GA] ensure there is language that gives precendence to
end over dur in case both are specified.

Tentative Resolution: [GA] default for timeAction is "display".

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Day 2 (Wednesday, September 15, 2004)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[GA] with help of [SH] formulates strawman rules for determining how
time activation and selection of hierarchical content into region
operates:

Rule 1: For an element to be selected into a region at time T, all of
its ancestors must be active at time T.

Rule 2: If an element is selected into a region, and any ancestor is
also selected into the same region, then only most senior such ancestor
A is combined into the region, with the active descendants of A being
candidates for presentation.

Rule 3: For an inline element to be combined into a region, its nearest
block element ancestor must be selected into same region.

[SH] Rule 3 means that it is meaningless to have a region attribute on
a span.

Action: [GA] remove region attribute from span element.

Resolution: do not support nested <div/> in DFXP at this time; if
compelling use-cases present themselves that are essential (meaning
that no other existing mechanism will satisfy need), then may reopen.
Reasons: (1) simplifies content and processing model; (2) encourages
authors to use AFXP for cases requiring richer model; (3) author can
use metadata elements to ascribe additional structure
independently;[N.B. This reverts the tentative resolution of
September 15 to support nested <div/>.]

Action: [GA], prompted by [SH], need to define what is meant by begin
attribute on <body>.

Action: [KB] lookup whether default time container for SMIL body is
seq or par. Report back by tomorrow (9/16).

Resolution: default time container value for <div/> is par.

Action: [GA] add table indicating default values for timeContainer.

Resolution: default time container value for <p/> and <span/> is par.

Resolution: do not support nested <span/> in DFXP at this time; if
compelling use-cases present themselves that are essential (meaning
that no other existing mechanism will satisfy need), then may reopen.
Reasons: (1) simplifies content and processing model; (2) encourages
authors to use AFXP for cases requiring richer model; (3) author can
use chained referential styling to ascribe nested style structure
independently; [N.B. This reverts the tentative resolution of
September 15 to support nested <span/>.]

Action: [GA] remove nested <span> support from schema and span
element definition.

Resolution: Allow multiple tts:* attributes on style element; serves
semantically like a group <g> and can alternatively be expressed using
<g>.

Action: [GA] Update schema to permit multiple tts:* attributes on
style element.

Action: [GA] make style attribute IDREFS instead of ID.

Action: [GA] add style attribute to <g/>.

[SH] Proposes to change region so that it uses style attribute to
refer to styles in addition to defining them inline as children.

Action: [GA] change region so that it uses style attribute to
refer to styles in addition to defining them inline as children.

Action: [GA] need to establish precedence for different style
association mechanisms as follows, with priority order as follows:
(1) inline, (2) referential or chained referential, (3) region
referential. In case that referential or chained referential results
in a diamond graph, then last instance of style in pre-order depth
first traversal wins.

[SH] there may be some details pertaining to Rule #2 adopted above,
the issue is when an element and an ancestor as selected into a
region, then does the styles of the region independently associate
with both ancestor and child, or does only the region of the ancestor
associat.

Action: [SH] will research issue of interaction of combination rule
#2 and region referential styling.

Action: [GA] add <p> and <span> to list of elements to which
background-color applies

Action: [GA] change initial value of color style property to
transparent. [N.B. consequence of this is that unless author
specifies a color, then content will be invisible.]

Action: [GA] restore "inherit" value for style properties.

Action: [GA] specify language about meaning of line height.

[DK] Need to have way to interoperably express coordinates in rows
and columns and to express total # of rows/cols for which content was
composed.

Action: [GA] formulate proposal to express coordinates in rows
and columns and to express total # of rows/cols for which content was
composed.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Day 3 (Thursday, September 16, 2004)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[KB] Reports that default time containment for <body> in SMIL is indeed
seq, not par.

Summary of Default Time Container Behavior

body	seq
div	par
p	par
span	par

[GA] Continuing with discussion of rows/cols for expressing lengths...

[GA] We will need way to express size of frame in which content was
composed in order to provide for transforms to other coordinate
spaces. For example, total # rows/cols. However, this applies for
pixels as well.

[DK] Need to have way to express frame rate in which content was
composed in case that time counts with "f" metric are used.

[GA] Confirms from [TM] that we need to use <meta>, not <metadata>.
See metadata draft module.

Action: [GA] change DFXP schema and WD to use <meta> instead of
<metadata>.

Example of EBU Teletext

<head>
  <meta id="ebu-teletext"
    ttm:frame-rate="25"
    ttm:frame-width="720px"
    ttm:frame-height="576px"
    ttm:pixel-aspect-ratio="4:3"
    ttm:content-width="40c"
    ttm:content-height="24r"
    ttm:content-x="72px"
    ttm:content-y="58px"
    ttm:horizontal-resolution="0.050r"
    ttm:vertical-resolution="0.011c"
  />
</head>

Action: [DK] Discuss with [MD] how to express pull-down parameters to
completely characterize frame and field composition.

Resolution: adopt following metadata items to describe composition
context: ttm:frame-rate ttm:frame-width ttm:frame-height
ttm:pixel-aspect-ratio ttm:content-width ttm:content-height
ttm:content-x ttm:content-y ttm:horizontal-resolution
ttm:vertical-resolution.

Action: [GA] need to add following metadata items to schema and spec:
ttm:frame-rate ttm:frame-width ttm:frame-height
ttm:pixel-aspect-ratio ttm:content-width ttm:content-height
ttm:content-x ttm:content-y ttm:horizontal-resolution
ttm:vertical-resolution.

Action: [GA] need to specify semantics for each metadata item, need
to specify what it means if multiple items with same name are present
in some context, and if not permitted, expected behavior.

Example of Foreign Metadata

<head>
  <meta xmlns:aml="http://www.bbc.co.uk/rd/aml">
    <aml:Program ProgID="W3CTT11A11" ProgTitle="Eastenders clip from
LDXG964E"/>
    <aml:Change date="2003-07-07 11:33:07" text="AML file created by
GUI." module="GUI:1.5.1w"/>
    <aml:Change date="2003-07-10 11:22:58" text="Script converted from:
W3CTT11A11.txt" module="convertScript:3.4"/>
    <aml:ShotChanges>
      <aml:Shot conf="0" type="cut" frame="349"/>
      <aml:Shot conf="0" type="cut" frame="385"/>
    </aml:ShotChanges>
    <aml:Settings>
      <aml:AssignColours colour="yellow" autoassign="yes"/>
      <aml:AssignColours colour="cyan" autoassign="yes"/>
      <aml:AssignColours colour="green" autoassign="yes"/>
      <aml:ReuseColours value="yes"/>
      <aml:OutputFormat value="EBU"/>
      <aml:SubtitleStyle style="bbc2" footers="yes" headers="yes"
justification="centre"/>
      <aml:Video file="\\pc479\video1\clips\ee_whoa.mpg"
audioearly="0.161" firstframe="900859" secretoffset="1"/>
    </aml:Settings>
  </meta>
</head>

Example of Agent, Actor Metadata

<head>
  <meta id="agents">
    <ttm:agent id="pierce_brosnan">
      <ttm:name>Pierce Brosnan</ttm:name>
    </ttm:agent>
    <ttm:agent id="richard_kiel">
      <ttm:name>Richard Kiel</ttm:name>
    </ttm:agent>
    <ttm:agent id="007">
      <ttm:name type="main">James Bond</ttm:name>
      <ttm:name type="alias">007</ttm:name>
      <ttm:name type="variant">0007</ttm:name>
      <ttm:actor ref="pierce_brosnan">
    </ttm:agent>
    <ttm:agent id="jaws">
      <ttm:name>Jaws</ttm:name>
      <ttm:actor ref="richard_kiel"/>
    </ttm:agent>
    <ttm:agent id="richard_kiel_as_himself">
      <ttm:name>Richard Kiel</ttm:name>
      <ttm:actor ref="richard_kiel"/>
    </ttm:agent>
  </meta>
</head>

Association #1

<p ttm:agent="007">Hurry, start the car!</p>
<p ttm:agent="jaws">Grrrrr!</p>
<p ttm:agent="richard_kiel_as_himself"/>Did I just crush that car?</p>

Association #2

<p>
<meta ttm:agent="007"/>
Hurry, start the car!
</p>

[DK] agent attribute type should be IDREFS, e.g., if more than one
agent is simultaneously producing referenced output.

Resolution: adopt following metadata items for use in DFXP:
ttm:agent, ttm:name, ttm:givenName, ttm:familyName, ttm:actor.

Action: [GA] add definitions of following metadata elements to schema
and spec for DFXP: ttm:agent, ttm:name, ttm:givenName,
ttm:familyName, ttm:actor. Also add ttm:agent attribute to <p> and
<span>.

[DK] Add following as well: ttm:title, ttm:description, ttm:copyright.

Action: [GA] add ttm:title, ttm:description, ttm:copyright to schema
and spec for DFXP.

Action: [GA] ensure that all elements can also take ttm:* attributes,
but need to define in context of specific elements which ones have
standard semantics.

Action: [GA] define semantics of ttm:agent metadata attribute such that
inner specifications override outer specifications, e.g.,

<p ttm:agent="lucy">
Don't you love me Ricky?
<span ttm:agent="ricky">No</span>
</p>

Resolution: change text-shadow to text-outline with value syntax as
follows: none | <color> <thickness> <blur-radius>, where thickness and
blur-radius are expressed as <length>.

Action: [GA] change text-shadow to text-outline with value syntax as
follows: none | <color> <thickness> <blur-radius>, where thickness and
blur-radius are expressed as <length>.

[SH] Add <span> to applies to for wrap-option, which allows overriding
wrap-option on paragraph for extent of spanned content.

Action: [GA] add span to wrap-option "applies to".

[GA] notes that we limit writing-mode to apply only to region in DFXP;
will generalize in AFXP to apply to content elements.

Action: [GA] for each AFXP style property that we don't include in DFXP,
need to specify a normative default value for DFXP, but do so in a way
that doesn't require normative use of AFXP specification.

Action: [GA] since <color>s are opaque, need to say so in verbiage.

[SH] points out problem with generic font families: monospace may be
serif or sans-serif, therefore there should be: monospace-serif and
monospace-sans-serif.

Action: [GA] Remove misspelled "monotype" entry from <generic-family-
name>, and replace it with "default". All other entries are also to be
removed for present time.

[GA] Notes existence of PANOSE proposal on W3C at [3]:

[3] http://www.w3.org/Printing/stevahn.html

[GA] We could request HP to propose use of PANOSE to the TT WG. Note
that this is not under consideration for DFXP, but should consider for
AFXP.

Action: [GA] Add row/col units to <length> style value expression.

Tentative Resolution: Keep both replace and append with combine attr.
Burden of proof on [GA] to define well.

Action: [GA] Ensure that if we define overflow="clip" on region that
we say that outermost region is expected to be clipped in all cases.

[SH] overflow should be a style property of region, and should have
following values from CSS: visible|hidden|inherit; initial value
of visible. Leave out auto and scroll for time being.

Action: [GA] change overflow attribute on region to use tts:overflow
style property, which applies only to region, and which supports values:
visible, hidden, inherit.

[SH] Strawman #1:

For normal clock values: 3h 2m 30.2s
For NPT equivalent media time values: 03:02:30:06 (@ 30fps)
For SMPTE marker values: smpte(03:20:03:00) == npt(03:02:30:06)

NPT values can be begin, end, or duration; while SMPTE marker values
can only be begin and end, but not duration.

Ambiguity

begin="03:02:30:06" (is this NPT or SMPTE marker?)

SMIL 2.0 Media Markers using SMPTE Codes as Markers

begin="foo.marker(smpte=03:02:30:06)"

[SH] proposes following shorthand for SMPTE time codes as marker for
begin/end (i.e., prefix smpte time value with '*'):

begin="*03:02:30:06"

[GA] Thinks there will be minimum use made of true SMPTE time values as
markers. More likely to use NPT media time values.

Resolution: Adopt following syntax for clock value:

clock-value
  : ( hours ":" )? minutes ":" seconds ( fraction | frames-and-subframes
)?
  | time-count fraction? metric?

frames-and-subframes
  : ":" frames ( "." subframes )?

frames
  : digit digit

subframes
  : digit digit

Action: [GA] Change clock-value syntax to support expressing optional
frames and subframes in order to express NPT media times. Define clock
value semantically to be NPT for TT media object. Remove optional "+"
value from offset-expression (in which case we don't need "offset-
expression" intermediate term).

Action: [GA] allow style to use style attribute to refer to other
styles, since we effectively now allow <style> to work like a group of
styles.

Action: [GA] remove "div" from display-align's applies to list. Can't
apply to div since div has no explicit height/width.

[DK] Reviewed (and tweaked) 8 style/region examples. See
20040916-ex1.txt at [4].

[4] http://lists.w3.org/Archives/Member/member-tt/2004Sep/0037.html

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Review Issues from Telecon of 2004-09-09
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[N.B. In the following, an asterisk (*) prepended to a paragraph
indicates that the paragraph came from the minutes of the 2004-09-09
telecon.]

*[DK] Should there be default values for each style defined in spec?

[GA] Yes. Defined in spec.

*[DK] Should there be default time metric?

[GA] Can author specify default? Is there a default default?

[GA] Propose <meta ttm:default-time-metric="f"/>, e.g., or
<meta ttm:default-time-metric="s"/>. Further, propose that if
author does not specify, then it be "s".

Action: [GA] Add ttm:default-time-metric metadata item. Takes
values that are one of time-count metric, e.g., "h", "m", "s",
"f", etc.

*[DK] Can a single style element specify multiple style attributes?

[GA] Yes, we have now permitted this.

*[DK] Should region be able to indirectly reference styling via "style"
attribute, as is the case with g and content elements?

[GA] Yes, we have now permitted this.

*[DK] Should there be both style attributes as well as style child
elements of region element?

[GA] This is permitted.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Review TT-Basic Requirements [5]
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[5]
http://lists.w3.org/Archives/Member/member-tt/2003Oct/att-0026/tt-af-1-0
-req_TT-Basic.html

[GA] Interesting potential additions to DFXP that were not excluded by
TT-Basic Requirements, and are not currenly in DFXP WD:

* hyperlinking (R216)
* embedded fonts (R219, R220)
* descriptive vocabulary (R221)
* scroll and hightlight animation (R502)
* animated styles (R505) - note this interacts with [DB]'s request to
  have timeAction="style"

Working Assumption: None of the above potentials additions will be in
DFXP unless there is a champion that advocates them and does all the
work of providing spec ready text (that meet's GA's standards).

[SH] quite opposed to including animated styling (R505 features).

[DK] will evaluate descriptive vocabulary for possible inclusion
in DFXP, possibly using existing metadata format.

[GA] N.B. Have not yet had serious review of [DB]'s request for
timeAction="style".

[GA] Following were not excluded by TT-Basic, but previous decisions
have made moot or WG now finds questionable in DFXP context:

* presentational flowed text vocabulary (R210, R215)

* out-of-line styling (R302, R303)

* the following visual style properties (from R306) 

  - absolute position
  - baseline alignment point
  - baseline alignment
  - baseline dominance
  - baseline shift
  - bidirectional treatment
  - block progression dimension
  - block scroll amount
  - border before, after, start, end
  - break before, after
  - clear
  - color profile name
  - display block, inline
  - float
  - indent start, end
  - inline progression dimension
  - inline scroll amount
  - intrusion-displace
  - line feed treatment
  - line stacking strategy
  - line wrapping option
  - opacity
  - padding before, after, start, end
  - reference orientation
  - relative position
  - space before, after, start, end
  - text altitude (ascent)
  - text decoration
  - text depth (descent)
  - text indent (first line)
  - text shadow -- we recast as text-outline for outline only function
  - white space collapse
  - white space treatment

* temporal fill mode (R307)

* style parameter symmetry (R390) - N/A due to removal of out-of-line
  styling (R302, R303); all style properties in DFXP are expressed
  as attributes effectively;

* out-of-line timing (R402)

* certain synchronization parameters (R403)

  - endsync
  - fill
  - fillDefault
  - repeatCount
  - repeatDur
  - restart
  - restartDefault

* certain synchronization parameter values (R404)

  - event values
  - access key values
  - wallclock values

Action: [GA] send John's document to WG reflector, then (at some point
in future) ask him to send to public list so it can be handled as public
submission.

Action: [GA] Add informative Annex to DFXP spec which enumerates the
requirements of TT-AF-1-0-REQ which are intentionally not intended to
be met in DFXP.

Action: [KB] Review the March 2002 TT Task Force requirements to see if
the current TT-AF-1-0-REQ does not include all those earlier
requirements,
and if not, then make a case for including in TT-AF-1-0-REQ. [TM] will
help as required and report back to WG.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Next F2F Meeting
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

[GA] Tentatively schedule next F2F for 11/30 - 12/02 in UK (most likely
Reading, MSFT hosting).

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
START SUMMARY
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

*** RESOLUTIONS ***

Resolution: retain <div/> in DFXP.

Tentative Resolution: support nested <div/> in DFXP.

Tentative Resolution: [GA] default for timeAction is "display".

Resolution: do not support nested <div/> in DFXP at this time; if
compelling use-cases present themselves that are essential (meaning
that no other existing mechanism will satisfy need), then may reopen.
Reasons: (1) simplifies content and processing model; (2) encourages
authors to use AFXP for cases requiring richer model; (3) author can
use metadata elements to ascribe additional structure
independently;[N.B. This reverts the tentative resolution of
September 15 to support nested <div/>.]

Resolution: default time container value for <div/> is par.

Resolution: default time container value for <p/> and <span/> is par.

Resolution: do not support nested <span/> in DFXP at this time; if
compelling use-cases present themselves that are essential (meaning
that no other existing mechanism will satisfy need), then may reopen.
Reasons: (1) simplifies content and processing model; (2) encourages
authors to use AFXP for cases requiring richer model; (3) author can
use chained referential styling to ascribe nested style structure
independently; [N.B. This reverts the tentative resolution of
September 15 to support nested <span/>.]

Resolution: Allow multiple tts:* attributes on style element; serves
semantically like a group <g> and can alternatively be expressed using
<g>.

Resolution: adopt following metadata items to describe composition
context: ttm:frame-rate ttm:frame-width ttm:frame-height
ttm:pixel-aspect-ratio ttm:content-width ttm:content-height
ttm:content-x ttm:content-y ttm:horizontal-resolution
ttm:vertical-resolution.

Resolution: adopt following metadata items for use in DFXP:
ttm:agent, ttm:name, ttm:givenName, ttm:familyName, ttm:actor.

Resolution: change text-shadow to text-outline with value syntax as
follows: none | <color> <thickness> <blur-radius>, where thickness and
blur-radius are expressed as <length>.

Tentative Resolution: Keep both replace and append with combine attr.
Burden of proof on [GA] to define well.

Resolution: Adopt following syntax for clock value:

clock-value
  : ( hours ":" )? minutes ":" seconds ( fraction | frames-and-subframes
)?
  | time-count fraction? metric?

frames-and-subframes
  : ":" frames ( "." subframes )?

frames
  : digit digit

subframes
  : digit digit

*** OPEN ACTION ITEMS ***

Action: [EH] Provide use cases in TT context for priority class attr
usage.

Action: [SH] To elaborate use cases for audio embed usage.

Action: [GA] to generalize and clearly articulate [MD] feedback on XBC
with TT-AF structured timing as use case, then send to WG for review
prior to forwarding to XBC.

Action: [GA] need to document conventions regarding normative definition
of element vocabulary and syntax used therein.

Action: [GA] need to characterize definitions of vocabulary as normative
while supplied schemas are informative.

Action: [GA] review recent proposed changes regarding xml:space from
XML Core Working Group with view of changing our vocabulary definitions
as necessary.

Action: [GA] restore nested <div/> in content model.

Action: [GA] ensure that background color applies to content elements
as well as region.

Action: [GA] change language in vocab element defs that defines equiv
class of attrs in foreign namespace to not admit unqualified attrs.

Action: [GA] need to ensure that we document whether time intervals
are half-opened intervals or closed. We always use half-opened intervals
for [begin,end) = [begin,begin+dur).

Action: [GA] ensure there is language that gives precendence to
end over dur in case both are specified.

Action: [GA] remove region attribute from span element.

Action: [GA], prompted by [SH], need to define what is meant by begin
attribute on <body>.

Action: [KB] lookup whether default time container for SMIL body is
seq or par. Report back by tomorrow (9/16).

Action: [GA] add table indicating default values for timeContainer.

Action: [GA] remove nested <span> support from schema and span
element definition.

Action: [GA] Update schema to permit multiple tts:* attributes on
style element.

Action: [GA] make style attribute IDREFS instead of ID.

Action: [GA] add style attribute to <g/>.

Action: [GA] change region so that it uses style attribute to
refer to styles in addition to defining them inline as children.

Action: [GA] need to establish precedence for different style
association mechanisms as follows, with priority order as follows:
(1) inline, (2) referential or chained referential, (3) region
referential. In case that referential or chained referential results
in a diamond graph, then last instance of style in pre-order depth
first traversal wins.

Action: [SH] will research issue of interaction of combination rule
#2 and region referential styling.

Action: [GA] add <p> and <span> to list of elements to which
background-color applies

Action: [GA] change initial value of color style property to
transparent. [N.B. consequence of this is that unless author
specifies a color, then content will be invisible.]

Action: [GA] restore "inherit" value for style properties.

Action: [GA] specify language about meaning of line height.

Action: [GA] formulate proposal to express coordinates in rows
and columns and to express total # of rows/cols for which content was
composed.

Action: [GA] change DFXP schema and WD to use <meta> instead of
<metadata>.

Action: [DK] Discuss with [MD] how to express pull-down parameters to
completely characterize frame and field composition.

Action: [GA] need to add following metadata items to schema and spec:
ttm:frame-rate ttm:frame-width ttm:frame-height
ttm:pixel-aspect-ratio ttm:content-width ttm:content-height
ttm:content-x ttm:content-y ttm:horizontal-resolution
ttm:vertical-resolution.

Action: [GA] need to specify semantics for each metadata item, need
to specify what it means if multiple items with same name are present
in some context, and if not permitted, expected behavior.

Action: [GA] add definitions of following metadata elements to schema
and spec for DFXP: ttm:agent, ttm:name, ttm:givenName,
ttm:familyName, ttm:actor. Also add ttm:agent attribute to <p> and
<span>.

Action: [GA] add ttm:title, ttm:description, ttm:copyright to schema
and spec for DFXP.

Action: [GA] ensure that all elements can also take ttm:* attributes,
but need to define in context of specific elements which ones have
standard semantics.

Action: [GA] define semantics of ttm:agent metadata attribute such that
inner specifications override outer specifications, e.g.,

Action: [GA] change text-shadow to text-outline with value syntax as
follows: none | <color> <thickness> <blur-radius>, where thickness and
blur-radius are expressed as <length>.

Action: [GA] add span to wrap-option "applies to".

Action: [GA] for each AFXP style property that we don't include in DFXP,
need to specify a normative default value for DFXP, but do so in a way
that doesn't require normative use of AFXP specification.

Action: [GA] since <color>s are opaque, need to say so in verbiage.

Action: [GA] Remove misspelled "monotype" entry from <generic-family-
name>, and replace it with "default". All other entries are also to be
removed for present time.

Action: [GA] Add row/col units to <length> style value expression.

Action: [GA] Ensure that if we define overflow="clip" on region that
we say that outermost region is expected to be clipped in all cases.

Action: [GA] change overflow attribute on region to use tts:overflow
style property, which applies only to region, and which supports values:
visible, hidden, inherit.

Action: [GA] Change clock-value syntax to support expressing optional
frames and subframes in order to express NPT media times. Define clock
value semantically to be NPT for TT media object. Remove optional "+"
value from offset-expression (in which case we don't need "offset-
expression" intermediate term).

Action: [GA] allow style to use style attribute to refer to other
styles, since we effectively now allow <style> to work like a group of
styles.

Action: [GA] remove "div" from display-align's applies to list. Can't
apply to div since div has no explicit height/width.

Action: [GA] Add ttm:default-time-metric metadata item. Takes
values that are one of time-count metric, e.g., "h", "m", "s",
"f", etc.

Action: [GA] send John's document to WG reflector, then (at some point
in future) ask him to send to public list so it can be handled as public
submission.

Action: [GA] Add informative Annex to DFXP spec which enumerates the
requirements of TT-AF-1-0-REQ which are intentionally not intended to
be met in DFXP.

Action: [KB] Review the March 2002 TT Task Force requirements to see if
the current TT-AF-1-0-REQ does not include all those earlier
requirements,
and if not, then make a case for including in TT-AF-1-0-REQ. [TM] will
help as required and report back to WG.

*** OPEN ISSUES ***

*** URIs ***

[1] http://lists.w3.org/Archives/Member/member-tt/2004Sep/0036.html
[2] http://lists.w3.org/Archives/Member/member-tt/2004Sep/0022.html
[3] http://www.w3.org/Printing/stevahn.html
[4] http://lists.w3.org/Archives/Member/member-tt/2004Sep/0037.html
[5]
http://lists.w3.org/Archives/Member/member-tt/2003Oct/att-0026/tt-af-1-0
-req_TT-Basic.html

*** NEXT MEETING DATES ***

Normal CC schedule.

2004 11/30 - 12/02 Reading, UK (tentative)

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
END SUMMARY
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Received on Tuesday, 19 October 2004 07:51:53 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 2 November 2009 22:41:29 GMT