See also: IRC log
<trackbot> Date: 02 October 2012
<johnsim-msft> 2. Previous meeting minutes on Sep 18 http://www.w3.org/2012/09/18-html-media-minutes.html
<scribe> ScribeNick: joesteele
<johnsim-msft> https://www.w3.org/html/wg/media/track/
<johnsim-msft> https://www.w3.org/Bugs/Public/show_bug.cgi?id=17682
johnsim-msft: propose a solution
to the clear key bug, did add a comment a few minutes ago
bug#17682
... some others we could discuss as well
<johnsim-msft> http://tinyurl.com/7tfambo
johnsim-msft: no analysis yet on closing these bugs
Title: HTML-Media
<scribe> Meeting: HTML Media Extensions
<johnsim-msft> https://www.w3.org/Bugs/Public/show_bug.cgi?id=17673 container guidelines https://www.w3.org/Bugs/Public/show_bug.cgi?id=17682 clear key https://www.w3.org/Bugs/Public/show_bug.cgi?id=19208 keymessage event not needed https://www.w3.org/Bugs/Public/show_bug.cgi?id=19156 switching decoders
johnsim: container guidelines
bug
... I had an action item on this.
... new bug coming on ISO based media file formats should have
stronger guidelines when a CDM has a key in it storage
ddorwin: no preference on order
johnsim: starting with container guidelines bug
… exchanged some email on this and added to the bug
<scribe> … new bug is 19208
johnsim: this is specific to the
COmmon Encryption Format specification
... please read the bug online
... this is the proposed language for the ISOBMFF
guidelines
… captures enough of how to detect encryption and extract the metadata needed from those files
… does raise some issues about initialization and events
… there are cases where the CDM does not need to issue a key message event
… any comments or concerns, David?
ddorwin: no comments
johnsim: next bug is bug#17682 (clear key bug)
… documents how keys and key ids are associated
<BobLund> mute
… added a brief comment to the end of the spec
<BobLund> mute me
johnsim: how clear key works
should be the same as the CDM
... should be defined by the W3C since this is strictly a W3C
concept
… data blob specific to CDM is in the spec, in the case of clear key just a key
… need to define how clearKey is initialized or any media format is supported
… propose some draft language based on some email traffic to be sent to general rflector
johnsim: need to add language for other formats
ddorwin: for ClearKey specific stuff, and ISO, nothing to define just a keyid
… for other containers need to define
… this is part#1 of the bug, part#2 is the license
… this is the harder part, need to define key/keyid pairing
johnsim: there have been some
proposals out there
... do you mean a defined data structure like a license being
passed down?
ddorwin: yes
… in the old version for WebM has a simple definition but this has been removed
… need a license to provide more than one key
markw: need to say how do we support more than one key at a time?
ddorwin: want clearKey to be as capable as other CDMs to use for prototyping
… maybe use JSON or something
johnsim: we are talking about the analog to a license for clearKey, requirement is that is be capable of communicating more than one key/keyID pair
… I am a little hazy on that
… init data would have to report more than one
ddorwin: nothing that says you can't have more than one keyid
johnsim: keyed is really a meta-key which tells the server to deliver more than one key at once
ddorwin: might need to be base64-encoded JSON but moves a little away from original spirit
johnsim: some work on this in the webCrypto group
<ddorwin> where the spirit was to be able to easily specify a key in JS
markw: you mean how keys are encoded? we just specify an arrayBuffer of the raw bytes at this point
johnsim: so this action item is
still open. Need to propose license structure for
clearKey
... should we move on?
... jumping to switching decoders bug #19156
… has been some discussion but no replies yet
ddorwin: unlike clearKey some CDMs will include decoders and/or rendering pipelines
… this means you may be picking a decoder by picking a CDM
… this may cause a switch in the decoder being used
… what should we do in that case?
… should we spec it away?
aaron: should not spec it away at this point
… could be implementation
johnsim: could have portions encrypted and some not
aaron: thinking of the case where content comes from multiple files
johnsim: the metadata would be from different files in this case?
aaron: starting with unencrypted file and moving to encrypted file, should not cause decoding to stop
ddorwin: need to specify the decoder up front
… how does media source handle this?
… currently only one key system allowed per media element
markw: currently decoder is not
allowed to change
... prefer not to prevent it for now if possible
... will comment on the bug
s: /markw: will comment/aaron: will comment/
aaron: as long as you are only switching during media segment boundaries, should avoid this problem
Aaron_Colwell: could have the CDM disable encryption when clear frames comes through
markw: you need the CDM to be able to handle this already
this is an issue for ad-insertion
<markw> my point was just that (i) the CDM should be able to cope with unencrypted samples and (ii) the first encrypted sample and all subsequent samples must not depend on any samples before the first encrypted sample
johnsim: so we will add comments to the thread on this?
ddorwin: will reply with what I heard
johnsim: moving to bug#19208
… based on the container guideline bug
… came across a case where the keyMessage event might not be needed
… David pointed out this contradicts the spec
… bug is -- if a session is created which activates a CDM, and key storage already has a key, no need for a keyMessage event
… David, what do you think?
ddorwin: does not violate the spirit of the spec
… but we might need to change the text of the algorithm (BMFF text)
johnsim: you don't disagree then
ddorwin: don't have any
objections, but would like feedback and thought
... calling createSession would not cause a message to the app,
which is a little strange
johnsim: had discussions about the event corresponding to a key successfully being extracted
… is the a need for that event (e.g. a null-key messages)
ddorwin: we have KEYERROR we could add a ALREADY_HAVE_A_LICENSE_ERROR
… createSession could return a reference to an existing session - need to think about that more
… would not matter so much whether you got another message then
johnsim: any other thoughts on
this issue?
... going back to the list of open bugs
<ddorwin> https://www.w3.org/Bugs/Public/show_bug.cgi?id=18531
johnsim: renaming addKey
<ddorwin> http://lists.w3.org/Archives/Public/public-html-media/2012Sep/0045.html
markw: there was a couple of
suggestions -- update() was the only one commented on
... would suggest we go with that
I like update()
ddorwin: only concern is that in the general case we are adding a license
johnsim: you are really adding things -- more than an update
markw: update() means providing more information
<ddorwin> http://dictionary.reference.com/browse/update
johnsim; connotation of providing entirely different information and adding values
<ddorwin> http://mw1.merriam-webster.com/dictionary/update
johnsim: there are quite a few open bugs
<ddorwin> transitive verb: to bring up to date
johnsim: what is the status for the open bugs?
<johnsim-msft> http://tinyurl.com/7tfambo
I did not get to the example I was to do.
https://www.w3.org/Bugs/Public/show_bug.cgi?id=16540
johnsim: are folks attending the TPAC meeting?
ddorwin: yes I will be
I am not going at this point
johnsim: encourage everyone on the call with open bugs to resolve or start a thread on these bugs ASAP
… want to reach closure at TPAC or before if possible
ddorwin: many are editorial
johnsim: would be great to pull out the substantial bugs
… design questions, etc
… any that sit at the top of the list as needs resolving?
ddorwin: the clearKey bug must be resolved
<adrianba> we don't have to close every bug - we need to ensure the document covers the desired scope before we ask for FPWD
ddorwin: needKey is another one
johnsim: which is the needKey?
ddorwin: will try to get the other bug I need to file out
… and send an email
ddorwin: no major changes to the spec in these bugs
johnsim: concern about the specs fitting together?
ddorwin: MSE is another source for the URL should work together, might need to discuss advanced use cases
<scribe> … new segments from other key systems, etc.
markw: will scribe for next meeting
ddorwin: TPAC is on the 29th
johnsim: next EME call will be
the last one before the TPAC
... would like to discuss how we can use the TPAC to advance
the spec on the next call
... would like to resolve as many bugs as possible before than
as well
... David the issues you raised we should really resolve
... raises our chances of coming out with a good draft
... pretty much done
bye all
This is scribe.perl Revision: 1.137 of Date: 2012/09/20 20:19:01 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/have to have/can't have/ Succeeded: s/meta keyed/meta-key/ Succeeded: s/er media/per media/ Succeeded: s/HAVE_A_KEY_ERROR/ALREADY_HAVE_A_LICENSE_ERROR/ Succeeded: s/aaron:/Aaron_Colwell:/ Found ScribeNick: joesteele Inferring Scribes: joesteele WARNING: No "Present: ... " found! Possibly Present: Aaron_Colwell BobLund CHAIRS Clarke FORCE HTML_WG IA_Team Microsoft MikeSmith P46 SPARQL SW_ SW_HCLS ScribeNick Team_Validat Title WAI_ WAI_UAWG WCAG2ICT XML_ET-TF aaaa aabb aacc aadd aaee aaff aagg aahh aaron active adrianba ddorwin glenn html-media https joesteele johnsim johnsim-msft joined markw matt pal strobe trackbot You can indicate people for the Present list like this: <dbooth> Present: dbooth jonathan mary <dbooth> Present+ amy WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Found Date: 02 Oct 2012 Guessing minutes URL: http://www.w3.org/2012/10/02-html-media-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]