Media Pipeline Task Force Teleconference

06 Oct 2011

See also: IRC log


John_Simmons, Franck_Denoual, Eric_Winkelman, Hiroyuki_Aizu, Clarke_Stevens, Jan_Lindquist, Duncan_Rowden, Kaz_Ashimura, Francois_Daoust, Mark_Watson, Mark_Vickers, Bob_Lund, Russell_Berkoff, Steven_Wright, Paul_Caporn


Clarke: got to requirement 2 last week. Let's move on from there.
... We can get through the rest of the less controversial ones today.
... Next week, we can get to adaptive streaming, then the week after that to the digital rights requirements.
... Then that's TPAC.
... Less concerned that we're not ready by TPAC. We want to show that we have requirements at TPAC. From the F2F, we do have general agreement about use cases and requirements even it not perfect.
... Main goal of TPAC is to bring these to WG.

John: is the intent to take all of them to TPAC or some of them?

Clarke: We only want to have standardized those things that can't be done today.

R3. Midstream Modification of Track Elements

<Clarke> R3: http://www.w3.org/2011/webtv/wiki/MPTF/MPTF_Requirements#R3._Midstream_Modification_of_Track_Elements

Clarke: ability to change, add, remove track element in the middle of a stream.
... From some discussion, maybe removing them is not necessary.

Bob: Changes are going into HTML5 spec as we speak to add this capability.
... Ian Hickson is leaving the door open to track removal if a use case comes for that.
... We agree that it's difficult to a) know when a track is gone, and b) do something useful with it.

markW: Potential problem if a track comes back?

Bob: It wouldn't get added as a new one.

markW: How would you know it's come back?

Bob: you'd look at cues.

markW: What about audio?
... [example with English track]
... Was planning to comment to HTML WG about that.

Russell: Similar comment, if you don't know when a track's gone, hard to see when a track comes back.
... The argument goes that in order for you to rewind, you need to have them available. Broadcast does not support rewind, so they should reflect what is available.

Bob: Didn't see anything in the bug's response that it was specifically about seeking.

Russell: maybe I read through the lines here.
... How do I have my default language, and over time want to switch to another default. If I don't know when a track is removed, I can't implement a selection mechanism based on preferences.

MarkW: valid points. Not quite sure what Ian proposes works in the end.
... What about track numbers when you remove a track?
... It has to be treated in a different way. A track that gets removed has to remain in the list.

Bob: I think that's right, otherwise you can't reorder track.

Russell: dealt with in OIPF, whether array of tracks is dynamic. Ended up with a static array parsed again and again.

Clarke: in the user agent, you then have to "track" which track maps to which index.

Russell: Yes.

Clarke: We don't need to change the requirement here necessarily. Checking that Jan and Mark's comments are recorded in that bug seems the appropriate way to go.

R4. Content Authorization Parameters

R4: http://www.w3.org/2011/webtv/wiki/MPTF/MPTF_Requirements#R4._Content_Authorization_Parameters

Clarke: Does this one have a bug?

Bob: No. It is my opinion that this can already be done.
... There are a bunch of other issues about parental issues that are not covered by the use case.

Clarke: could be covered by the requirement 2 on being able to distinguish types of metadata

R2: http://www.w3.org/2011/webtv/wiki/MPTF/MPTF_Requirements#R2._Key_Metadata_Types

John: I took an action item to identify ways to document metadata conventions. Two options inside W3C for this kind of documentation: community and business groups.

Community and Business Groups

John: this could be the way to get together and work on this kind of specification.

Clarke: Do people agree that a community/business group would be the place to work on metadata mapping?

Bob: I'll be able to share a few things with this group.

R5. Content Authorization Failure

R5: http://www.w3.org/2011/webtv/wiki/MPTF/MPTF_Requirements#R5._Content_Authorization_Failure

Clarke: There are a number of reasons that could explain why you cannot play a content.

MarkW: This seems to be a narrow example of a more generic problem for an error reporting mechanism.
... It's a real requirement, but I wonder if we could report on the generic error reporting mechanism and enumerate the examples we have here as part of it.

Clarke: It may just be the addition of an error, or if the error reporting mechanism is not efficient enough.

R6. Adaptive Bit Rate Format Support

R6: http://www.w3.org/2011/webtv/wiki/MPTF/MPTF_Requirements#R6._Adaptive_Bit_Rate_Format_Support

MarkW: question here is what do we want to normatively specify here? That different solutions be supported?

Bob: We had a couple of specific requests (going through)

Clarke: we have to be able to tell when it is an adaptive streaming format.

<mav> I don't think R6 is a gap

<mav> Suggest delete R6 & leave R7 & R8

MarkW: We need to be clear what we're requesting. If it is something that can already be satisified with no changes, it doesn't make sense to list that as a requirement.

JanL: I'm having the same comment as Mark. Could we flag the requirement as needed but nothing needs to be done?

MarkW: If there had been a RF codec, then they would have made a requirement. The same applies for adaptive streaming format. If we can have DASH be that format, then good. Unclear we can, though.

MarkV: I think it's going to evolve in the next couple of years. Even if we have a RF format, I don't think we should put it forward yet.

John: HTML WG is being format agnostic to let the market decide which format to use.

Clarke: I'm not seeing any strong push to keep that requirement in there.
... I propose to leave it for a few days. If somebody has some strong argument, please make them. Otherwise, I propose to drop it.

R7. Adaptive Bit Rate Parameters (and R8. Adaptive Bit Rate Feedback)

R7: http://www.w3.org/2011/webtv/wiki/MPTF/MPTF_Requirements#R7._Adaptive_Bit_Rate_Parameters

Clarke: Requirement we captured at the F2F.

JanL: I don't believe there's any mechanism today, e.g. to specify the max.

Clarke: Could we say we have a preference of a way to do this?

JanL: My preference would be to have a dialog with WG who will pick up this requirement.
... I don't have a solution. Many times, you come to W3C with a proposal. Here I don't have that, so that's why I suggest a dialog.

MarkV: R7 and R8 are the counterparts of R4 and R5. Parameters in the media pipeline, one way for R4 and R7, or the other way for R5 and R8

JanL: I note we could change "failure" into "feedback", as we need a generic feedback mechanism, not only for failures. I agree with MarkV that it's a good parallel.

Clarke: suggest to merge them?

JanL: no.

MarkV: I agree on keeping them separate, just adjust the wording to use same words.

Clarke: ok, so we covered R8 as well. Going on to security.

R9. Security and Digital Rights Management Identification

R9: http://www.w3.org/2011/webtv/wiki/MPTF/MPTF_Requirements#R9._Security_and_Digital_Rights_Management_Identification

Clarke: Parallel with R6. Is the conclusion the same?
... For instance, if I am to use UltraViolet, can I do it already?

Bob: I don't know for UltraViolet. I think that, in general, anything can be encoded as mime type parameters.

Clarke: Do we have anyone from DECE who knows what we might need?

John: DECE does define a very specific encoding format, not just down to the container, also pixel details.
... Encoding works fine with downloading and adaptive streaming.
... Both DASH on-demand and live demand can be encrypted for instance. Not DRM specific.
... It's a part 12 compliant encoding (ISO base file format).
... I don't see additional requirement for HTML
... You may want to know whether the device supports a given encryption mechanism.

JanL: I was a bit confused as to how it worked in a streaming world where you keep sending authentication information.

John: you can have information in the MPD telling if it's DASH, where to get the key.
... It's just not a file.
... In terms of key rotation and things of that nature, that's also supported in adaptive streaming as well.
... [example of things to do on the production side]
... From a browser's point of view, there are lots of reasons why you will fail to play content, in particular while dealing with one of the DRM subsystems.
... Even if you may be able to require a key, you may still end up not being able to play the content.

Clarke: Two basic questions. R9, I haven't heard anything here that says we cannot do it today.
... Second, I hear more stuff to expose parameters in R10.

Bob: Second point, similar issue as adaptive bitrate, but different information that needs to be conveyed.

Clarke: Are there some additional security functionalities that need to be added beyond being able to expose parameters and provide feedback?

MarkW: [scribe missed that]

John: [missed beginning] It's a piece of piece process. You can pick up one, such as device identification we're talking right now.

JanL: I just sent a bug proposal based on Hollywood meeting.
... If people can comment. ISSUE-18. I'll pick it up next week in any case.

John: We have to be very concise in the requirements for 9 and 10. For example, DECE Utlraviolet is too vague, because you might have different systems that support different functionalities.

JanL: Points I raised previously, retrieval of ? and device identification, are these precise enough?

John: I think so.
... For some, it's more a matter of user experience issue, where you don't want to discover you cannot play after having tried.

JanL: As I presented in Hollywood, this is something Open IPTV Forum touched upon.

Clarke: Any work that we can reference is useful, yes.
... I'm going to suggest that R6 and R9 are already covered, and then I'll try to parallelize R10/R11 with R7/R8 and R4/R5.
... Thanks everyone, be active online and talk to you next week!

[Call adjourned]

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.136 (CVS log)
$Date: 2011/10/07 12:55:38 $