{minutes} HTML WG media telecon 2013-05-14 - MSE and EME bug discussions

The minutes are available at:
http://www.w3.org/2013/05/14-html-media-minutes.html
________________________________

[http://www.w3.org/Icons/w3c_home]<http://www.w3.org/>

HTML Media Task Force Teleconference
14 May 2013

Agenda<http://lists.w3.org/Archives/Public/public-html-media/2013May/0082.html>

See also: IRC log<http://www.w3.org/2013/05/14-html-media-irc>

Attendees

Present
Bin_Hu, BobLund, Cyril, Michael_Thornburgh, acolwell, adrianba, ddorwin, glenn, jdsmith, joesteele, markw, pal, paulc, pladd
Regrets

John Simmons

Chair
Paul Cotton
Scribe
joesteele
Contents

 *   Topics<http://www.w3.org/2013/05/14-html-media-minutes.html#agenda>
    *   Roll call<http://www.w3.org/2013/05/14-html-media-minutes.html#item01>
    *   Previous Minutes<http://www.w3.org/2013/05/14-html-media-minutes.html#item02>
    *   Action items and issues<http://www.w3.org/2013/05/14-html-media-minutes.html#item03>
    *   ACTION 11<http://www.w3.org/2013/05/14-html-media-minutes.html#item04>
    *   Open Issues<http://www.w3.org/2013/05/14-html-media-minutes.html#item05>
    *   MSE status and bugs<http://www.w3.org/2013/05/14-html-media-minutes.html#item06>
    *   Arrange pre-last call period for MSE<http://www.w3.org/2013/05/14-html-media-minutes.html#item07>
    *   EME bugs<http://www.w3.org/2013/05/14-html-media-minutes.html#item08>
    *   FPWD<http://www.w3.org/2013/05/14-html-media-minutes.html#item09>
 *   Summary of Action Items<http://www.w3.org/2013/05/14-html-media-minutes.html#ActionSummary>

________________________________

Roll call
Previous Minutes
http://www.w3.org/2013/04/30-html-media-minutes.html http://www.w3.org/2013/05/07-html-media-minutes.html
<paulc> VER 2 agenda: http://lists.w3.org/Archives/Public/public-html-media/2013May/0082.html
paulc: using later version of the agenda
Action items and issues
paulc: 2 items and one issue
... ACTION 10 was on Adrian
adrianba: there's an action for everyone to arrive at consensus on model for the session
... have not caught up on the status lately
paulc: bug 19208 is still open?
adrianba: may be
paulc: resolved as a dup of 21855
<paulc> Bug 21855 - Avoid network traffic and duplicate sessions for the same key(s)
<paulc> https://www.w3.org/Bugs/Public/show_bug.cgi?id=21855
ACTION 11
paulc: write a proposal for the JS -- Mark did this and the link is in the agenda
<adrianba> ACTION-11?
<trackbot> ACTION-11 -- Mark Watson to write a proposal for the case where the data is not available to the JS -- due 2013-04-16 -- OPEN
<trackbot> http://www.w3.org/html/wg/media/track/actions/11
paulc: believe this is now done?
markw: yes
paulc: marking as closed
ddorwin: showing as closed now
Open Issues
paulc: this is re: moving the ClearKey spec into a separate docs
<paulc> https://www.w3.org/html/wg/media/track/issues/1
MSE status and bugs
paulc: updated a couple of the sub agenda items based on Aarons status report
... Aaron has updated the editors draft which is in the agenda
... sent a message giving the update information
... Aaron do you want to add?
acolwell: no - just working on editorial changes
<paulc> Aarron's update on MSE: http://lists.w3.org/Archives/Public/public-html-media/2013May/0075.html
acolwell: will provide some additional text this week
paulc: any comments on this?
... assume that only outstanding bug it the splicing bug?
<paulc> Bug 21431 - Specify splicing behavior for text tracks
<paulc> https://www.w3.org/Bugs/Public/show_bug.cgi?id=21431
paulc: so you will impleent what is in comment #12?
<paulc> Planning to implement: https://www.w3.org/Bugs/Public/show_bug.cgi?id=21431#c12
jdsmith: at the F2F talking about video quality we settled returning dropped frames as the indicator of quality
... it is actually difficult to measure dropped frames
... more accurate for us to measure rendered frames
<adrianba> https://www.w3.org/Bugs/Public/show_bug.cgi?id=20760
jdsmith: original proposal was to return average frame rate of rendered frames
... returning dropped frames is problematic in our implementation
paulc: bug number?
<paulc> Current status of 20760 is RESOLVED FIXED
acolwell: question is that we should have the total number and rendered frames, so should calc difference
jdsmith: we have approximate average frame rate so we can approximate
<paulc> [Paul apologizes for his rough cough if it is audible]
acolwell: this should be accurate?
paulc: are you asking to reconsider?
jdsmith: yes -- would prefer to return the data points and let the app figure it out
paulc: would prefer you reopen with explanation, possibly with revised text
jdsmith: ok
acolwell: not clear what the interval is that the frame rate is being measured on -- wary of that metric
markw: re: dropped frames should be simple as possible for the user to understand
... don't want something simple for us but harder for users?
... approximation is fine for this purpose
pal: wanted to follow up on the issue of supporting multiple frame rates
... is this really a requirement?
acolwell: yes
pal: so why not limit every buffer to a single framerate?
acolwell: we don't guarantee a seamless splice between buffers
paulc: do we have a separate bug for this question?
acolwell: no - this discussion came up during time splice accuracy discussion
<markw> In H.264/ISO media files at least there is no such thing as frame rate. Each sample has a timestamp, that is all
paulc: if this is orthogonal - having a separate bug with a clear disposition would be a good paper trail
... leave it to others to decide
acolwell: WebM also
paulc: Jerry you will take the action to reopen as necessary
cyril: question about the text track splicing
<paulc> Bug 21431 - Specify splicing behavior for text tracks
cyril: have not read carefully, but was something said about text tracks having variable frame rates?
... may not know the duration of the cue
... until the next track comes in
... might have some impact on the management of the cues
acolwell: we are assuming we know the duration of cues now
... may be a requirement we need to state -- need text track experts to chime in
cyril: for example video survellance vidoes may not know the frame rate until the vides come in
... may extend the duration of the previous sample
acolwell: is this a common use case we need to solve now? not all MPEG-4 file supported now
... can file a bug to discuss more
cyril: we can resolve later, but should recognize the issues
acolwell: some discussion has been around defining simple behavior and adding more complex logic later to describe the outlier cases
cyril: I will open a bug
BobLund: comment on the idea of a cue with unknown length
... not really unknown from a media buffer perspective
... it will have an end, but player may not know the end
... HTML5 is clear on what should happen with the rendering of text tracks
pal: confirm current thoughts on behavior
... for purposes of MSE, in the algorithm frames are replaced by cues
acolwell: non-overlapping cues
... lots of nastiness when overlapping cues are allowed
paulc: any other bugs for MSE we should discuss?
cyril: have seen in the latest spec we say
... the MOOV is not allowed in a media segment
... will this trigger a decode error when encountered?
... spec says what is conformant but not non-conformant
acolwell: let me look at the spec
<acolwell> https://dvcs.w3.org/hg/html-media/raw-file/default/media-source/media-source.html#sourcebuffer-segment-parser-loop
acolwell: look at step 2 of the parsing loop
<paulc> Quote: 2.If the input buffer starts with bytes that violate the byte stream format specifications, then run the end of stream algorithm with the error parameter set to "decode" and abort this algorithm.
acolwell: this shows the signal of a decode error
cyril: when you have a MUST and it is not fulfilled then it is a violation
acolwell: yes
... if you have some text that would make it clearer please suggest it
cyril: just in a different location, but I understand now
paulc: cross-references in the spec to make it clearer are useful
... should be useful for 3rd parties who are not in these discussions
cyril: so maybe a cross-ref to the byte stream specification would help
acolwell: add a normative note
Arrange pre-last call period for MSE
paulc: this will close Wed May 22nd -- this email went out a couple of weeks ago
... we must know what bugs to process prior to Last Call
cyril: if we want bugs resolved before must be filed by then?
paulc: no -- any bugs filed before that date will be resolved before Last Call
... otherwise it prevents us from going into Last Call
... because we are not at 0 bugs
paulc: this lets us identify the bugs that must really be addressed by Last Call
... moving on to EME?
cyril: what is the status of the primer?
paulc: did a call for volunteers -- have not received any volunteers as yet
<paulc> Call for volunteers: http://lists.w3.org/Archives/Public/public-html-admin/2013May/0002.html
cyril: might want to volunteer
paulc: might want to have a conversation with the chairs to confirm what is wanted
... get some context on previous primers
EME bugs
http://dvcs.w3.org/hg/html-media/raw-file/tip/encrypted-media/encrypted-media.html
paulc: David want to make any comments?
ddorwin: no -- sent an email
<paulc> https://dvcs.w3.org/hg/html-media/raw-file/tip/encrypted-media/encrypted-media.html
FPWD
<paulc> FPWD: http://www.w3.org/TR/2013/WD-encrypted-media-20130510/
<ddorwin> Also, http://www.w3.org/TR/encrypted-media/
paulc: need to update the copyright to match the editors draft
Topic Outstanding Bugs
paulc: 30 of them
<paulc> http://tinyurl.com/7tfambo
paulc: there are is an active email thread on ACTION-11 -- should we spend time on that? or avoid it?
<paulc> http://lists.w3.org/Archives/Public/public-html-media/2013May/0068.html
<markw_> I think it makes sense to talk about the ACTION-11 item
ddorwin: marks email since there has been discussion
paulc: intro to ACTION-11?
markw: proposal was to define what might happen if CDM data is not available to the JS apis
... we know there will be CDMs that takes advantage of HW and data will not be available
... may be cases where it will be available
... need to define behavior when the data is not available
... two cases - audio and video
... Canvas could be used to access video data with getImageData to get raw YUV or RGBa data from Canvas
... with audio there is a Media Source Element source mode
... can do the processing for the audio node
... get back the raw PCM bytes
... could have the original APIs behave as if there was no media -- probably most likely to be implemented
... drawImage would do nothing, Media Element source mode would act as if there is no audio
... could image another scenario where they work but just not for the EME data
<pal> Q: shouldn't the JS application be able to inform the user that video/audio is not available because it is protected?
markw: would be much more complicated
... maybe we should not exclude this possibility
... maybe we just need to document this behavior?
adrianba: since we won't be able to definitely describe the behavior for all possible APIs now and in the future for EME
... we should make this informative following the principle Mark outlined
... would act as if no data was available
... maybe we can use the Canvas and audio as examples
... let APIs define there own behavior based on this principle
<markw_> What adrian says makes sense to me
<ddorwin> +1
markw: that is question about the failure mode
... that would allow the JS to respond differently
<adrianba> You can detect in JS that you're using EME since you have to do something explicit to use it.
markw: something the owners of both APIS to specify
... app should know if it is dealing with encrypted media
agree with the proposal -- however could be an issue with when some streams are encrypted and others are not, for example ads.
<adrianba> i think we can wait to see if apps run into problems with not knowing whether failure was because of EME or otherwise before engineering a solution to what seems to me to be a corner case
pal: to marks comment - the existnece of EME does not necessairly imply this, seems to be a reluctance to directly indicate this
markw: the way to indicate if it is to be indicated it specific to the APIs (Canvas, audio)
pal: the Canvas element has errors, conceivably we could file a bug that another error is needed e.g. CANNOT_COPY
markw: could be fine to file a bug, up to them whether to expose this
pal: more information better for the JS app
adrianba: echo what Mark said, hard to enumerate all the places where this data might be consumed.
... this seems like an edge case as to why the data was encrypted
... unlikely that an app will have something the user can correct to handle this
... if that emerges we can solve the problem then
paulc: what do we need to do to decrease the bug count?
... do we need to nominate a specific list of bugs?
adrianba: bugs are waiting for feedback -- does not need to be in this meeting
... half of the bugs have proposed solutions
... should not be too difficult but folks need to reply to them
<ddorwin> Most recently updated bugs: https://www.w3.org/Bugs/Public/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&columnlist=short_desc%2Cassigned_to%2Cpriority%2Cbug_status%2Cresolution%2Ctarget_milestone%2Cchangeddate&component=Encrypted%20Media%20Extensions&product=HTML%20WG&query_format=advanced&list_id=11029
adrianba: most recent bugs would be a good target
... could have a conversation then
<ddorwin> Sorry, this might be better: https://www.w3.org/Bugs/Public/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&columnlist=short_desc%2Cassigned_to%2Cpriority%2Cbug_status%2Cresolution%2Ctarget_milestone%2Cchangeddate&component=Encrypted%20Media%20Extensions&list_id=11029&product=HTML%20WG&query_format=advanced&order=changeddate%20DESC%2Cbug_status%2Cpriority%2Cassigned_to%2Cbug_id&query_based_on=
paulc: you are saying we need folks input before discussing in next weeks meeting
... could be more aggressive in proposing changes to force folks forward
<ddorwin> Previous email listing of bugs to discuss: http://lists.w3.org/Archives/Public/public-html-media/2013May/0004.html
paulc: resolving bugs as NEEDS_INFO
... nearly out of time -- please suggest a list of bugs and people whose input is needed
<ddorwin> New bug: https://www.w3.org/Bugs/Public/show_bug.cgi?id=21977
paulc: will have a mtg next week starting with EME and could start assigning folks to make progress
adrianba: filed a new bug about swapping parameters
... people should take a look
<ddorwin> https://www.w3.org/Bugs/Public/show_bug.cgi?id=21977 - Swap isTypeSupported() parameter order so keySystem is first
adrianba: link is pasted above
paulc: done for today
paulc: propose focusing on EME next week and MSE after
paulc: done for today
Summary of Action Items
[End of minutes]
________________________________
Minutes formatted by David Booth's scribe.perl<http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm> version 1.138 (CVS log<http://dev.w3.org/cvsweb/2002/scribe/>)
$Date: 2013-05-14 16:06:43 $