W3C home > Mailing lists > Public > public-html-media@w3.org > September 2012

minutes for MSE call 28 Aug 2012

From: Matt Womer <mdw@w3.org>
Date: Tue, 4 Sep 2012 10:19:54 -0400
Message-Id: <B677C690-8499-4353-9135-A32243AD01B9@w3.org>
To: "public-html-media@w3.org" <public-html-media@w3.org>
Hi all,

Joe Steele took these minutes, but unfortunately RRSAgent wasn't around to grab them.  Thanks to the logs from Joe and Adrian, I've recreated them here:

And as text below.




      [1] http://www.w3.org/

                               - DRAFT -

                  HTML Media MSE weekly teleconference

28 Aug 2012


      [2] http://lists.w3.org/Archives/Public/public-html-media/2012Aug/0049.html

   See also: [3]IRC log

      [3] http://www.w3.org/2012/08/28-html-media-irc


          NiXu, pierre, joesteele, Matt_Womer, PaulC, BobLund,
          acolwell, markw, Clarke, duncanr, ddorwin, Matt,
          adrianba, paulc, boblund, Nixu, kstreeter, pal, johnsim,





     * [4]Topics
         1. [5]Role Call
         2. [6]Minutes from last meeting
         3. [7]Review of Tracker Action Items
         4. [8]Baseline Docs and bugzilla
         5. [9]Update to the MSE specs since previous meeting
         6. [10]Actions from previous meeting
     * [11]Summary of Action Items

   <paulc> paulc, who is here?

Role Call

   <scribe> Scribe: joesteele

Minutes from last meeting

   paulc: any comments on the meetings? action items have ben
   carried forward

Review of Tracker Action Items

   paulc: there are none for Media Source

Baseline Docs and bugzilla

   paulc: noted

Update to the MSE specs since previous meeting

   paulc: three parts to Aarons update messages
   ... 13 bugs have been resolved by these updates
   ... let's not go through them all unless Aaron wants to
   highlight something or more discussion needed

   acolwell: not anything in particular - mostly editorial or

   paulc: are all marked as fixed?

   acolwell: yes

   paulc: anyone want to comment/discuss more on these?
   ... moving on ..

Actions from previous meeting

   paulc: 7 actions items -- let's check the status
   ... bug#18400
   ... Define and document timestamps heuristics -- reply to MarkW
   response by the next meeting

   https: //www.w3.org/Bugs/Public/show_bug.cgi?id=18400

   paulc: ball is back in Aaron's court?

   acolwell: yes

   <paulc> Mark did his action. See

     [12] https://www.w3.org/Bugs/Public/show_bug.cgi?id=18400#c5

   paulc: next item - bug#17000

   johnsimm: need more research - still pending

   paulc: bug#17002 -- specificy source id to video track id
   ... Aaron posted a message -- need followup?

   acolwell: is this sufficient? if so then can update the spec

   adrianba: did not get all the way through
   ... where does the original id come from? is there a way to
   create it?

   <paulc> Aaron's update is at

     [13] https://www.w3.org/Bugs/Public/show_bug.cgi?id=17002#c7

   acolwell: no depends on the original expectation - no specific
   guidance given by the spec as long as unique

   adrianba: my reading is could come from the source element

   <paulc> See also Aaron's email to the list at

     [14] http://lists.w3.org/Archives/Public/public-html-media/2012Aug/0008.html

   adrianba: could come from the media file itself
   ... if it was coming from the media, reason to believe the
   application could know the mapping from the media
   ... if created programmatically, not clear how the mapping is

   <paulc> Possible bug on Media Fragments WG work:

     [15] https://www.w3.org/Bugs/Public/show_bug.cgi?id=17002#c5

   acolwell: 2 issues -- correlation to a source buffer,
   correslation to a media segment
   ... issue with media segment, may have different track ids as
   could change during playback
   ... are we ok with the track id changing during playback? what
   should we do?
   ... current spec allows this to change

   adrianba: are there two issues? which source buffer is related
   to a specific track and how to get there from a particular

   acolwell: would be good to separate out these
   ... now that object-oriented model exists

   paulc: Silvia sent a comment to media wg -- is this another

   acolwell: there is a disconnect between media frag spec and the
   HTML5 spec in the language
   ... also text track does not have an ID attribute like video
   and audio tracks do
   ... there is no way to get an ID for a text track even with
   this change

   paulc: this sounds like 2 sep bugs on the HTML5 spec
   ... Silvia may send this to the media frag WG

   <scribe> ACTION: acolwell file a bug on HTML5 spec that text
   tracks should have an id like audio and video tracks do
   [recorded in

   acolwell: should a file a bug on how to handle?

   <scribe> ACTION: acolwell file a bug to specify how the ID per
   track is generated with the Media Source Extensions [recorded

   paulc: moving on
   ... bug#17006


     [18] https://www.w3.org/Bugs/Public/show_bug.cgi?id=17006

   paulc: Aaron you had an actton on this

   acolwell: did not do this yet

   <paulc> We need someone with more DASH experience to take this

   markw: I can take that ACTION 17006

   paulc: next bug#17094


     [19] https://www.w3.org/Bugs/Public/show_bug.cgi?id=17094

   paulc: Bob Lund was the champion

   BobLund: yes I have done that -- in the process of pulling
   together the proposed text. Should be ready in the next few
   ... discuss in the next few weeks

   paulc: please send email to the list
   ... next bug#16998

   https: //www.w3.org/Bugs/Public/show_bug.cgi?id=16998

   paulc: Aaron was going to split the bug

   acolwell: have not done that yet -- but lots of other stuff

   paulc: moving to next agenda item
   ... 23 bugs outstanding

   <adrianba> [20]http://tinyurl.com/6pdnzej

     [20] http://tinyurl.com/6pdnzej

   paulc: lots of bugs have come in from Opera

   acolwell: yes -- bugs are high quality though

   paulc: most of these bugs are clarifcations

   acolwell: stuff is not specific enough or looks normative but
   is not

   paulc: bug#180601 -- have we discussed?

   https: //www.w3.org/Bugs/Public/show_bug.cgi?id=18601

   Contradictory requirements for initialization segments

   acolwell: spec is inconsistent about whether init segments are
   ... lot of behavior is centered around this
   ... optional language makes this confusing
   ... should we still say they are optional?
   ... all current formats have init sgements
   ... this would be to support formats that come later which do
   not have them

   paulc: can you give us a ptr to the most recent spec?

   johnsim: is the issue that some media might not require init

   acolwell: believe Mark wanted the optional text but cannot
   think of one today that has that
   ... Mark is copied on the bug

   paulc: assign an action item to Mark on this

   acolwell: Bob -- for Transport stream stuff you still have this

   BobLund: yes

   <matt> trackbot, status?

   <scribe> ACTION: markw review the proposal in bug#18601
   [recorded in

   <trackbot> Sorry, couldn't find user - markw


     [22] https://www.w3.org/Bugs/Public/show_bug.cgi?id=18575

   paulc: next item

   acolwell: most of section 2 looks non-normative
   ... suggested to mark the whole section as non-normative
   ... I suggest changing to be normative instead
   ... should split bug into multiple sections to change to
   ... or keep as a single bug?

   paulc: there are 11 sub-sections
   ... what was original intent? descriptive text for reader and
   normative text for implementor?
   ... Aaron you are suggesting converting all the text to


   paulc: would be better to move the normative stuff out of
   section 2?

   acolwell: not sure

   <markw> there is a lot of material in section 2 that should be

   paulc: suggest you do one or two of these changes and come back
   to the group

   adrianba: lot of things that are normative in this section -
   should get an idea of what to do as we change it

   <markw> users of the API need to be sure how the different
   append cases will be handled, for example

   paulc: maybe change a couple of good examples on section 2 and
   come back to group with the examples as a proposal

   acolwell: some pieces are conceptual and some are more
   algorithmic - hard to see what is the right balance

   paulc: no problem with having both types, but make sure it is
   clear which is which

   <scribe> ACTION: acolwell to give a couple of examples for
   section 2 [recorded in

   <trackbot> Created ACTION-6 - Give a couple of examples for
   section 2 [on Aaron Colwell - due 2012-09-04].

   paulc: moving on to next


     [24] https://www.w3.org/Bugs/Public/show_bug.cgi?id=18642

   acolwell: wanted people to look at this bug and get feedback
   ... on the behavior specified
   ... would like someone with more experience with this type of
   edge case to respond

   paulc: anyone with this type of feedback?

   acolwell: live is probably the problem

   johnsim: Aaron if you could send me an email with a synopsis I
   can look for an answer

   acolwell: Bob or Duncan could you chime in as well?

   <Clarke> I'll take a look also

   <duncanr> ok will look at it too

   acolwell: this was in reference to transport streams

   paulc: next item



     [25] https://www.w3.org/Bugs/Public/show_bug.cgi?id=18708

   acolwell: current spec says when you call end of stream can't
   add any more data
   ... might be nice to append after this if a higher quality
   segment has been found later
   ... this bug lifts the restriction, if you append transitions
   back to open state
   ... is this useful?

   paulc: any comments?
   ... next is bug#18709


     [26] https://www.w3.org/Bugs/Public/show_bug.cgi?id=18709

   acolwell: adding a method to source buffer allowing you to
   flush data out of the source buffer
   ... one of the ideas was to allow app more control over how
   much data buffered in user agent
   ... also used to signal parts of timeline which are not
   important anymore

   <paulc> See response in

     [27] https://www.w3.org/Bugs/Public/show_bug.cgi?id=18709#c1

   ???1: what is the assumed behavior if it can act on its own?


     [28] http://dvcs.w3.org/hg/html-media/raw-file/tip/media-source/media-source.html#source-buffer-segment-eviction

   acolwell: basically the user agent can do what it needs with
   regard to releasing memory but tries to avoid stopping playback

   johnsim: allows app to give the UA a pass on playback for a
   piece of content

   acolwell: allows UA to not have to guess

   <markw> sometimes there are contractual restrictions on how
   much content can be buffered and this gives the application
   control of that

   acolwell: agreed with the comments from Phillip

   adrianba: think you said this allows app to be proscriptive,
   but earlier seemed like a hint and allowed the UA more

   <markw> I think it should be prescriptive

   acolwell: it should be prescriptive

   adrianba: is this intended to be synchronous then?

   acolwell: intended to be like an overlap

   paulc: think that handles Aarons suggested list
   ... some of the other bugs may be processed by him as well
   since they will be covered by his other work

   acolwell: will keep attacking clarification bugs first and new
   features second

   paulc: lots of actions
   ... meet again on Sept 11th in two weeks
   ... scribe volunteers?
   ... any other comments?

   adrian will send matt the notes from this meeting

   paulc: meeting is adjounred

Summary of Action Items

   [NEW] ACTION: acolwell file a bug on HTML5 spec that text
   tracks should have an id like audio and video tracks do
   [recorded in
   [NEW] ACTION: acolwell file a bug to specify how the ID per
   track is generated with the Media Source Extensions [recorded
   [NEW] ACTION: acolwell to give a couple of examples for section
   2 [recorded in
   [NEW] ACTION: markw review the proposal in bug#18601 [recorded

   [End of minutes]

    Minutes formatted by David Booth's [33]scribe.perl version
    1.133 ([34]CVS log)
    $Date: 2012/09/04 14:17:09 $

     [33] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
     [34] http://dev.w3.org/cvsweb/2002/scribe/

Scribe.perl diagnostic output

   [Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.133  of Date: 2008/01/18 18:48:51
Check for newer version at [35]http://dev.w3.org/cvsweb/~checkout~/2002/

     [35] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Mirc_Timestamped_Log_Format (score 0.89)

Succeeded: s/priveuous/previous/
Succeeded: s/till pending/still pending/
Succeeded: s/lot's/lots/
Succeeded: s/mark:/markw:/
Succeeded: s/ar enot/are not/
Succeeded: s/pass playback/pass on playback/
Succeeded: s/???1/johnsim/
Found Scribe: joesteele
Inferring ScribeNick: joesteele
Default Present: NiXu, pierre, joesteele, Matt_Womer, PaulC, BobLund, ac
olwell, markw, Clarke, duncanr, ddorwin, Matt
Present: NiXu pierre joesteele Matt_Womer PaulC BobLund acolwell markw C
larke duncanr ddorwin Matt adrianba paulc boblund Nixu kstreeter pal joh
nsim matt
Agenda: [36]http://lists.w3.org/Archives/Public/public-html-media/2012Au
Got date from IRC log name: 28 Aug 2012
Guessing minutes URL: [37]http://www.w3.org/2012/08/28-html-media-minute
People with action items: acolwell markw

     [36] http://lists.w3.org/Archives/Public/public-html-media/2012Aug/0049.html
     [37] http://www.w3.org/2012/08/28-html-media-minutes.html

   End of [38]scribe.perl diagnostic output]

     [38] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm
Received on Tuesday, 4 September 2012 14:20:05 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:32:56 UTC