Web&TV IG f2f meeting in Santa Clara

27 Oct 2014


See also: IRC log


(This section will be provided later.)
Yosuke, Mark, Giuseppe
Kaz, Francois, Daniel, jcverdie, Alex, giuseppe, yosuke


<kaz> Scribe: Kaz

<scribe> scribenick: kaz

Welcome and Logistics

yosuke: good morning
... welcome to the meeting
... this is a one-day meeting
... including some joint meetings
... if you have any ideas for today's meeting, you can put them
... myself has changed my affiliation to the W3C Team
... Kaz is taking notes
... would start with brief introduction

ddavis: Daniel Davis from W3C/Keio
... Team Contact

jfoliot: John Foliot

raphael: chair of media fragments

skim13: Sung-Hei

<scribe> (continues...)

yosuke:We need volunteers for scribing.

tidoust: volunteers to scribe for the second session

ddavis: for 11:00-13:00

yosuke: need 4 more scribes

alex: for 14:00-15:00

jcverdie: for 15-15:30

yosuke: for 17-18
... tx all

IG Introduction - Giuseppe

(slides to be added)

giuseppep: [Goals]
... look at not specs but big pictures of requirements
... media-centric apps
... identify gaps
... review of deliverables of other groups
... give feedback
... also liaison with other organizations
... [Web and TV IG]
... IG != WG
... link to resources
... wiki and home page
... phone calls and email discussions
... [Web and TV Interest Group Work Flow]
... figures out our work
... extract requirements from use cases
... and then gap analysis
... highlight real gaps
... with the existing specs
... sometimes need new APIs
... or sometimes need a new area
... may create a new Community Group
... [IG work overview]
... Closed TFs: Home Network, Media Pipeline, Testing, TImed Text

Giuseppe: [Home Network TF (closed)]
... contributed to the DAP WG
... privacy, security, etc.
... [Media Pipeline TF (closed)]
... adaptive streaming and content protection
... Gap analysis went to HTML5
... HTML5 getting a REC tomorrow
... [Testing TF (closed)]
... open web platform testing

yosuke: We had an initial report
... of testing from W3C at the Boston AC
... meeting this year: Some part of the activity worked
... really fine and we learned lessons as well.
... With continuing successufl part, currently
... thinking about next plan

giuseppep: [Timed Text TF (closed)]
... TTML vs WebVTT
... asked W3C to create only one format
... also mapping from one to another
... accepted by the TTWG
... [Media APIs TF (ongoing)]
... previous work: metadata, etc.
... current work: requirements for audio fingerprinting/watermarking

yosuke: comments/questions?


Agenda Bashing

mark: agenda posted for today
... want to make sure it works for everybody
... a lot are covered
... first, some work on work with other SDOs
... talk about what they're working on
... then
... break and
... a couple of joint sessions with other W3C groups
... TV Control API CG spun out from the IG
... Second Screen Presentation CG
... HTML WG Accessibility TF
... then lunch
... and back to the work of the IG: new area
... SMPTE and GGIE work
... and update report of key features, e.g., HTML5 EME, MSE and in-band resource issues
... related the work in the IG
... afternoon break
... and back to new use cases of the IG
... then wrap-up and next steps
... anything to add?

chris: is 15min enough for second screen?

mark: the first session with the TV Control API CG is discussion
... but second screen/accessibility should be brief update

louay: 15 min presentation and ... 10 min Q&A seems fine.

jfoliot: regarding the accessibility tf
... brief update today, and another joint session on Thursday

mark: will add 10 more mins for second screen
... we can be flexible

Updates from SDOs

<tidoust> scribe: Francois

<tidoust> scribenick: tidoust

Giridhar_Mandyam: [ATSC 3.0 update]
... working from Qualcomm, update for ATSC perspective
... 3 years ago, ATSC board approved work on ATSC 3.0
... no requirement to be backwards-compatible
... Redesign of the ATSC transmission top to bottom. Looking at physical layer, codecs. Another major change is looking at new transport layer, investigating alternatives to MPEG2-TS.
... One thing that is most important to ATSC for CE manufacturers and broadcasters is hybrid services.
... Organization of ATSC 3.0 in different groups, inclusing S31 (requirements), S32 (physical layer), S33, and S34 (apps and presentation) which is the most relevant for this group.
... The process flow: we started with a list of about 150 requirements. Some of them overlapped, of course.
... We created a bunch of sub-groups, e.g. audio.
... We decided to go to IP transport instead of MPEG2-TS. Two proposals have come in for broadcast purpose: ROUTE/DASH, using a FLUTE derivative, and MMT/DASH, using an MPEG standard.
... Both alternatives are supposedly compatible with DASH.
... They both leverage ISO BMPFF file container. A few differences, compatible with W3C MSE in theory.
... It very much looks like MPEG DASH will be the delivery mechanism for broadband.
... DASH.js open-source client will be compatible with these proposals, which is good news for W3C, since that leverages many Web technologies
... Hybrid content is also useful for supplemental content, including non real-time content.
... Getting to the runtime environment: initial decision to leverage HbbTV 2.0 as basis. Some possibility to constrain or extend that.
... Some liaison with HbbTV. HbbTV 2.0 is still MPEG2-TS centric.
... Input from Web and TV IG is being considered. Thanks for putting together that liaison letter!
... Some other areas related to W3C: program announcement and service guide to leverage OMA BCAST 1.0.
... Watermarking is in-band transmission of information that doesn't disturb the audio/video streams, but enable the client to extract that data and provide additional services based on that.
... No decision as to how to do that in a Web-based runtime environment yet.
... We've also been looking at emergency alerts, no decision yet either.

Dong-Young:Watermarking, is it proprietary or standardized?

Giridhar_Mandyam: A bit of both. Some parts may be standardized, others won't be because companies have their core business based on that. ... Future will tell how much can be standardized.

Mark_Vickers: [DLNA update]
... DLNA has been working on CVP2 for some time, now renamed to VidiPath. Presenting slides from Amol Bhagwat, from CableLabs and DLNA board.
... Works with any service provider for video. If the service provider has some gateway (e.g. set-top box), the idea is to use that gateway to connect all different devices available and deliver the video to all users devices.
... If TV supports this, the TV will do discovery across the network, and discovers which gateway(s) it can use.
... The Gateway gives the TV a URL to the program that it can interact with. The guide is HTML5. At this point on, the TV is running an HTML5 application.
... As opposed to the approach taken in the Home Networking TF, the TV does discovery initially, and then just runs HTML5 content.
... Requirements are diverse, 2 sets of requirements in practice: Application model is fully based on W3C, IETF, and related Web standards.
... We believe that the TV Web should be part of the Web. No non-standard Web API in the spec!
... Then there are a lot of requirements on the user agent, including support for MPEG-DASH, media formats, discovery protocols.
... All these things are implemented by the underlying user agent.
... There's a 1.1 version that is coming along for cloud guidelines, to be completed next month, on DRM and MSE, as these things are becoming more stabilized.
... A certification program was launched this September.
... Certified products expected to hit the market in December.
... CableLabs is doing a lot of work towards certification. In addition to that, they are also providing a full open source reference software stack.
... Now, looking at it from a Comcast perspective: From 2011 Web and TV workshop keynote, there were a lot of efforts

... done to embrace and extend existing specs.[Showing the list]

UNKNOWN_SPEAKER: Too many of them.

Mark_Vickers: HTML5 came along and did not use any of them.
... We thought it wise to abandon all these references and fully embrace the Web platform.
... Comcast efforts are to use HTML5 as defined by W3C, without extension. For instance, CEA-2014 was in the first DLNA RUI, but is not in VidiPath, to ensure that people would stop using it.
... If we find new APIs that are missing, we push for the work to be addressed here. ... Ongoing works on reference design kit (RDK).

ddorwin: Is it a goal that browsers can run VidiPath?

Mark: Browsers can decide to go through certification process. Yes. We want to make these things available to browsers.

Giridhar_Mandyam: Lots of discussion on Network Service Discovery API within this work. Assuming it's not part of RDK since Chromium expressed concerns on privacy/security issues. Your point of view?

Mark_Vickers: APIs that expose servers in the local network for me is a problem. Every company in the world can offer Web services on your local network, assuming that the home network is somewhat safe, because it is protected by a firewall
... One way of exposing network services for me is to do it at the chrome level. That's safer.
... Same as "window.print". The app does not even know if you're going to use a real printer in the end, how many printers there are in your network.
... Another example is photographs, using a dedicated DLNA server. If you give permissions to an external site to your DLNA server, you have no way to ensure that the external site will not pull out all the content.
... I feel it's too dangerous.

Jean-Claude_Dufourd: So, basically, we're looking at an API that says: "give me a photo server" and you get a pointer to it without knowing anything about its location or inner properties.

<ddorwin> The statement "We want to make these things available to browsers." was actually Comcast's position on the content, not necessarily via VidiPath.

Jean-Claude_Dufourd: I hear you want something like the Presentation API.

Mark_Vickers: Yes, I think, that's a very good example.

<ddorwin> There are a variety of reasons that it's very unlikely browsers would support VidiPath.

Alex: Is VidiPath constrained to the home network?

Mark_Vickers: DLNA has always been scoped to home networks, so that's the main focus, yes. But you can imagine using the cloud. I don't know if the organization is going to define that part though.

<ddorwin> Providers supporting VidiPath will likely also need a pure W3C-based solution to reach browsers on both desktop and mobile.

Giuseppe_Pascale: [HbbTV update]
... HbbTV stands for Hybrid Broadcast Broadband TV.
... Started as a French/Germany initiative but is starting to be adopted in many different countries.
... Most EU countries are using, trying or considering using HbbTV. Also deployed in Australia, announced in Russia, etc.
... Looking at the specification, it's really a list of references, from ISO, OIPF, and W3C of course.
... In HbbTV2, main new technology is the update of the Web profile used within the specification, based on HTML5, reusing the profile defined by the Open IPTV Forum. Profile just means that there is a minimum subset of specs that need to be supported by all HbbTV devices.
... The only legacy left from CEA-2014 is around the video. The <video> element from HTML5 can be used, but the old <object> API is still around, partly for backward-compatibility and for broadcast video.
... Other areas that are specific in HbbTV include "DRM in a CAM (Conditional Access Module)" through DVB CI Plus 1.4, HEVC, subtitles using TTML as profiled by the EBU, recommandations to handle privacy (Do Not Track support), and multi-stream synchronization where specific APIs were defined.
... Companion screens scenarios are supported: partially proprietary and partially standardized. Addresses ways for the TV to launch an app on a companion device and the opposite way around.
... App2App communication is a way for two apps to talk to each other on two different devices. A WebSocket server is running on the TV. Used as a bridge between the apps.
... To launch an app remotely on the TV, the DIAL protocol (Netflix/Youtube) is used. There have been lots of discussions on security and privacy considerations.
... Another major area of development is synchronization of applications and content across devices. Use cases include sign-language or second camera within the same device or across devices.
... Finally solution leverages DVB protocols, withing the TM-CSS group, using the App2App mechanism to signal synchronization mechanism.
... Main improvements on MPEG DASH. Some problem with Advert insertion into VoD content, using two HTML5 media elements with additional requirements for smooth transition between them.
... Also Push VoD.
... Some minor improvements on a few other areas.
... A strong emphasis on testing. The HbbTV v2.0 spec is basically ready but won't be published before a list of assertions is assembled.
... In v1, the attempt to get members to contribute tests did not work out so well. In v2, money from the HbbTV Association is being used to invite members to contribute tests.
... As said, the spec is ready, just missing a few test assertions.
... No major work on next version for the time being, as focus in on v2. One area of research is focus on IPTV.
... HbbTV is not going to do an "HbbTV IPTV profile" but that's not needed.
... An area of interest is Independent Application Signaling. Some networks filter the signaling. There are some discussions on how to work around that. Watermarking and fingerprinting could be a solution. That's one possible way of achieving this.

Giridhar_Mandyam: Going back to DRM on CAM slide. This was a little confusing for me. HbbTV claims to be compatible with HTML5 video but that does not seem to be compatible with the way DRM works.

Gisueppe_Pascale: Not my area of expertise. It's leveraging previous work in OIPF.

Giridhar_Mandyam: So my point is that it's not entirely compatible with HTML5 MSE.

Giuseppe_Pascale: Right. When this was specified, HTML5 MSE and EME were not stable enough. Might change in the future.

Mark_Vickers: General issue for SDOs on how to reference on-going W3C specs. We had to go to the board of DLNA to make an exception to be able to reference HTML5 since it's only becoming final now.
... Implementations are key here.
... Otherwise you just end up being too far behind.

Giuseppe_Pascale: Yes. It's a bit related to certification as well.

Mark_Vickers: For certification, we're leveraging as much as possible tests from existing W3C test suites. They are not enough from a certification perspective, but usually good.

Dong-Young: Support for MSE?

Giuseppe_Pascale: MSE is not part of HbbTV v2. For ad insertion, HbbTV relies on two video elements, used in sequence. All under the control of the application. Not a specific API per se, just guidelines to do it.

Jean-Claude_Dufourd: It seems the HbbTV Association does not trust discovery. The initial pairing of device is entirely proprietary. A bit weird from my perspective.
... Before the TV knows your tablet, you have to have some proprietary stuff going on to pair the devices.

Giuseppe_Pascale: Right, but needed for security reasons.

Jean-Claude_Dufourd: Also, companion screens is optional, kind of weird given the hot focus on that.

Giuseppe_Pascale: Yes, that was a compromise.

Yosuke_Funahashi: Any description about buffer size to enable synchronization between broadcasting signals and content from the Internet?

Giuseppe_Pascale: I'm not so familiar on that topic.

Yosuke_Funahashi: Any liaison between HbbTV and DLNA on certification?

Giuseppe_Pascale: There are conversations going on, trying to see if joint work can be possible. Nothing to report at this stage.

Hisayuki_Ohmata: [Hybridcast update]
... Hybridcast is a platform for broadcast and broadband hybrid service. It has been already launched in Japan.
... Talking about updates from TPAC last year: only NHK provided Hybridcast services, then trial services by 16 commercial broadcasters and, recently, TBS (Tokyo Broadcasting system) launched.
... Specification is now version 2.0
... In version 2.0, three major topics to enhance broadcasters' service: time-shifting viewing, supporting the video element for VoD with associated APIs for PVR, then companion device interface and accurate synchronized presentation.
... Also expansion of platform for service providers.
... Going into more details on each topic.
... For time-shifted viewing: MSE and EME are supported, video content of MP4 and MPEG-DASH can be played. APIs for playback control are also defined.
... For companion device interface: App to app communication, using native app and HTML5 app interfaces are defined.
... For synchronized presentation: one of the important factors, we think. [showing a video demo]
... IPTV Forum Japan defines the architecture model for synchronization and also APIs for controlling synchronized playback and exposing the broadcast timestamps.
... There are two types of Hybridcast applications: broadcast-oriented app and non broadcast-oriented app.
... The first ones are launched from the broadcast signal. The second ones are typically launched by the user.
... From the standpoint of services, non broadcast-oriented apps can
... provide cross-channel services, but not broadcast apps.
... Different considerations for security and content integrity: apps are executable under the permission of broadcasters and they can control API access.

Louay_Bassbouss: What do you mean by multiple apps may run at the same time?

Hisayuki_Ohmata: Several apps can run on the same screen at the same time, meaning multiple widgets can run in parallel. [multiple browsing contexts]

Mark_Vickers: Could you make available the extension APIs that you came up with? Send them to the IG so that we can look at them for instance?

Hisayuki_Ohmata: Version 1 is published. Version 2 has not been translated to English yet.

Mark_Vickers: It would be useful to see the extension APIs, not the common ones. It would be good to know those in these groups, as they could be used to extract requirements for standardization of new APIs.
... I'd like to make the request.

Hisayuki_Ohmata: I'll see what I can do.

Yosuke_Funahashi: Version 2 is published, the English translation is missing, it should be doable.

Donghoon_Lee: [TTAK.KO-07.0111/R1 "HTML5 Based Smart TV Platform" update]
... Working TTAK Technology of Telecommunications in Korea. ICT development, Certification and testing services in Korea.
... HTML5 based smart TV Platform is a specification for running Smart TV app, based on HTML5. No dependency on broadcasting system.
... The goal is to promote Smart TV Echo platform system.
... TV and set-top box devices can receive broadcast signals, and non linear streams and data through broadband.
... Looking at the architecture, the HTML5 based smart TV platform is built on top of TV operation system, split into app specific functions, Web core, and device specific functionalities.
... JavaScript-based APIs.
... The work on the spec started in March 2013 and first spec was published in March 2013. In April 2014, a second spec wxas published (revision 1) that adds second screen, adaptive streaming in particular.
... Current work to enable new features such as payment, second screen and content sync.
... Now looking at the application model.
... Four criteria to classify Smart TV application: 1) execution method (whether it comes from the Store, started by broadcasters signaling or launched by user from the broadband)
... 2) app packaging (packaged or not).
... 3) Broadcasting relation (activated or inactivated). Some apps may be suspended.
... 4) Channel Bound, depending on what happens when user switches to another channel.
... On the app lifecycle, an app can be launched by the user, an API call, or can be automatically started. An app can be terminated when the channel changes, through an API call, etc.
... Some W3C specification are not suitable to TV environment, so profiling has to be done. Based on OIPF profile.
... In addition, some Korean requirements on remote controller for virtual KeyCode definition. Geolocation API is included, media APIs profile for broadcast video.
... Extended API: such as access to broadcasting resources. These specs are not defined by W3C, so have to do it ourselves.
... [showing extended API structure tree]
... The extended API are exposed on the "window" object through "window.tvExt". Different interfaces.
... The app interface to control the current application, and those that are installed on the device.
... The broadcast interface to get info about TV channel and program. Can give the URL, name, description, and different info on program.
... The device interface to get detailed info on the device itself, including make and model.
... The Multiscreen & DRM interfaces are optional in this specification.
... No extended API to control broadcast video and channel operation, instead we adopted HTML5 video element. We defined new semantic requirements when used for broadcasting video and channel.
... Two states for broadcasting video: binding state and unbinding state. Setting/Resetting the src attribute on the video switches from one state to the other.
... In the binding state, video is controlled by the application, controlled by TV device otherwise.
... [presenting a code example with HTML5 and JavaScript code]
... To support app signaling, the spec references the DVB signal spec.
... On app packaging, the spec defers to W3C packaging format, introducing new URI scheme "tvapp", and adding the "apptype" and "appdomain" elements.
... [presenting demos]
... [First demo shows application lifecycle]
... [Second demo shows an example of multiscreen service application]
... There's another specification for conformance tests for HTML5 based Smart TV platform.
... It defines a test environment and test case.
... We also developed test dev framework, shared with everyone.
... I presented the Smart TV Platform, and certification program. Thank you very much.

<Zakim> kaz, you wanted to ask how to use geolocation api and to ask about the notation for sourceURI and to ask if these APIs are already available LG/Samsung TVs

Kaz_Ashimura: A few questions. You mentioned Geolocation API. How do you use that API?

Donghoon_Lee: W3C Profile for Smart TV, subset of W3C specifications.

Yosuke_Funahashi: Ok, but what's the use case?

Donghoon_Lee: In case of Smart TV, it detects the location based on the gateway, and looking for some nearby stores for instance.

Kaz_Ashimura: Thank you. You showed some notation about sourceURI. Was wondering about that.

Donghoon_Lee: Unique URL to identify the channel.
... Unique identifier for the channel.

Kaz_Ashimura: Is this spec already implemented in LG or other TV sets?

Donghoon_Lee: Currently, in case of Samsung and LG, own private applications.
... Broadcasters preparing to start these services based on this specification.


yosuke: We'd like to move second screen and accessibility media TF topics to after lunch.

TV Control API Community Group

<ddavis> scribenick: ddavis

<scribe> scribe: Daniel

yosuke: There are many efforts in W3C for convergence of TV and web standards.
... One result of that is the TV Control API Community Group. The chairs for that are Bin Hu and JC Verdie

<raphael> Will all presentations be linked from http://www.w3.org/2011/webtv/wiki/Face-to-face_meeting_during_TPAC_2014?

Bin: The TV Control API CG was spun off from the TV IG based on gap analysis work last year.
... We're started the group work in May to look into controlling TV tuners.
... We've done some use cases, requirements and gap analysis.
... We're looking at TV control APIs from Mozilla, Webinos, other SDOs as well as Samsung and LG.
... We want to create a candidate baseline and then further that work for a specification.
... Are there any suggestions for other agenda items?


Bin:First, let's review action items: http://www.w3.org/community/tvapi/track/actions/open
... Action 11 for yosuke is closed ( http://www.w3.org/community/tvapi/track/actions/11 )
... Action 14 for kaz is closed ( http://www.w3.org/community/tvapi/track/actions/14 )
... Action 16 is closed ( http://www.w3.org/community/tvapi/track/actions/16 )
... Action 17 is closed ( http://www.w3.org/community/tvapi/track/actions/17 )
... Next is an overview of specs in Japan.

tomoyuki: Here's an overview of Cable STB APIs in Japan
... In Japan there are three kinds of broadcast service types: terrestrial, satellite and cable
... TVs in Japan can playback terrestrial and satellite without STBs but for cable a STB is needed.
... In Japanese broadcast standards, there are two types of application runtime.
... One is BML (a subset of HTML 4.01), the other is a cable-specific STB API.
... Our cable-specific API is limited to additional definitions related to ARIB
... Three years ago, Japan Cable Labs (JLabs) defined this specification
... Two years ago, it was standardized as an ITU-T Recommendation J.296
... One is a JavaScript API, the other is a Java API (mostly ported from JavaScript API)
... It's mostly equivalent to OIPF-DAE
... You can obtain program information, recording, tuning, etc.
... One extension is to support Japanese broadcasting standards such as ARIB and JLabs cable standard.
... The Java API is almost completed imported from DAE with almost the same functionality as OIPF-DAE.
... The main difference is Android-specific support - intent support and search using a content provider.
... The spec was defined three years ago so the current spec is different to the current OIPF-DAE v. 2.3
... Our spec is based on OIPF-DAE v. 2.1
... We don't have an update yet because we have to watch the trend of common APIs for broadcasting.
... This is why we joined these Web and TV activities.

Bin: Thanks to Tomoyuki.
... Do you think the spectrum of those specs can fit some of the work we're doing in the TV Control API CG?
... Maybe you can tell us later when we look at the gap analysis. ... Mapping page: https://www.w3.org/community/tvapi/wiki/Main_Page/Requirements_Mapping

ddavis: We'll also add the public TV API from Samsung once it's been updated soon. ... TV Control API CG wiki page: https://www.w3.org/community/tvapi/wiki/Main_Page

Bin: We have about 80 technical requirements in the wiki page.
... The work now is to identify which features are used and which functions/classes are used by other APIs.
... We can then see which functions are the best fit for a baseline spec.
... There are general tuner requirements.

Tuner requirements: https://www.w3.org/community/tvapi/wiki/Main_Page/Technical_Requirement#General_Tuner_Requirements

[Bin is reading the tuner requirements]

Bin: The Mozilla API supports most of the requirements.

giuseppe: What was the goal of this analysis?

Bin: To look at what other specs are and to establish a baseline spec for this group.
... For areas that are not supported in other groups, we can fill those gaps.

Mark_Vickers: Is it the goal to support every one of those pieces of functionality?

Bin: Yes, that's the goal.

Mark_Vickers: My sense is that this is problematic. We should be developing APIs that work across all platforms - PCs, mobile, whatever.
... E.g. a webcam sending out an IP stream should be dealt with them in the same way.
... The current analysis here seems very hardware-specific.
... It doesn't seem to be bringing the TV world into an IP world. It seems to be creating a tuner API for the old world and excludes modern channels.

Bin: Not all of this is hardware-specific.

Mark_Vickers: If you look at a webcam, it has an IP address on the internet. There should be a unified model for dealing with traditional and non-traditional channels.

giuseppe: We want to reduce what's in HTML5. We should start from HTML5 and see what's the delta that's missing, instead of trying to adapt an old API.

gmandyam: Have you looked through the individual APIs and decided what should be exposed and what should be left as proprietary APIs?

Bin: What we're looking at is from the TV industry perspective.
... and looking at what will be run in the web environment.

gmandyam: For example, looking at the Mozilla API, there's an unlock parental control feature.
... From ATSC point of view, we would not want to expose that for developers to control.
... So what should we expose to developers?

Bin: The requirement now is that input from the TV industry is that that is a requirement and we didn't have any disagreements about that.
... Maybe we should go back and revise the requirements.

yosuke: Have you looked at the API from the BBC, I mean, TAL, TV Abstraction Layer?

Bin: No

yosuke: BBC created a library to abstract the API over SmartTV APIs from various CD manufactures
... - it's available on GitHub and so has become a kind of standard.
... [FYI: TAL: http://fmtvp.github.io/tal/getting-started/introducing-tal.html]

jcverdie: Last time I heard it was more of a prototype and at a higher level, like an application facility more than solving middleware control like we're trying to achieve here.

Chris_Needham: The TAL is at a higher level than this really. It provides a set of widgets and components .
... You let the TV app layer generate the markup for you - UI presentation.

giuseppe: The starting point was using the web to control the TV. If you look at apps running on a channel the requirements could be different.
... Maybe grouping them all together could create confusion.

Bin: The angle of this work is from a Web OS perspective.
... The application perspective would use those APIs as well as presentation features.
... It's a good idea to investigate it from a different perspective.
... It would be helpful if you could follow the work and we could open the work on the requirements.
... So we could have two perspectives - one is for applications, one is for the platform.
... If the angle is a platform/OS one then we should highlight it.
... So I'll highlight that.

<scribe> ACTION: Bin to update the gap analysis to indicate what is at an OS level. [recorded in http://www.w3.org/2014/10/27-webtv-minutes.html#action01]

<trackbot> Created ACTION-210 - Update the gap analysis to indicate what is at an os level. [on Bin Hu - due 2014-11-03].

Dong-Young: Regarding the system-level APIs, they'll be used by device makers themselves so they don't really need to be open to developers.
... They'll be interested more in application-level APIs so it would be more valuable to concentrate on thoses.

Bin: I think standardizing developer APIs would ultimately benefit the consumers.

Dong-Young: I'm not saying we shouldn't standardize system-level APIs, I'm saying we should prioritize application-level APIs.

Bin: Thank you. If you're able to look through the APIs on the wiki page, it would be helpful if you could highlight what you think is higher priority.
... I would encourage people to edit the mapping wiki page ( https://www.w3.org/community/tvapi/wiki/Main_Page/Requirements_Mapping )
... Please add enhancements and things like priorities.
... The group has a conference call every four weeks which we'll continue.

ddavis: Is the plan to look at requirements again?

Bin: We're going to concentrate on the mapping analysis but people can contribute to the requirements.

ddavis: If there are people here from organizations that have a TV API, we'd love to be able to use them for reference.

<tidoust> [Mark mentioned the possibility to treat channels as a "playlist" of live streams. That reminds me of abandoned work on a Gallery API worked upon by the DAP WG: See http://dev.w3.org/2009/dap/gallery/ ]

ddavis: Please either add your API details to the wiki or just let us know the URL and someone from the group will do the mapping.

Bin: Thanks everyone and to the chairs for letting me speak.

Second Screen Working Group

Second Screen Community Group website: http://www.w3.org/community/webscreens/

Second Screen Presentation API: http://www.w3.org/2014/secondscreen/presentation-api/20140721/

Oops - this one is more recent...

Second Screen proposed API: https://webscreens.github.io/presentation-api/

Louay_Bassbouss: I'm going to show this using the second screen API itself.
... I have the slides in a browser on my mobile device
... If I click a button, I get a dialog asking which application I want to present.
... I can then send the slides to the presentation device.
... So this is an actual use case of this API.
... We have about 62 participants in the second screen presentation community group.
... There are several technologies to enable second screens.
... The goal for us is to show web content on a second screen.
... Web content means not just web pages but also video, etc.
... In this example we've used a mobile device which is the controller page.
... It uses the requestSession(url) function to send content to the second screen (presenting page).
... The user agent allows the user to select the second display.
... User interaction is always required for security.
... There are many technologies that can be used for this - DLNA, Mifi, Miracast, etc.
... The difference between some of these technologies is that sometimes you don't have two user agents - maybe just one with frames sent to the second display.
... The Presentation API abstracts from these technologies.
... Gaming and media flinging are other use cases.
... The current status - in July we published the final report.
... This month the Second Screen Presentation Working Group was created.
... We currently have 19 participants from 10 organizations.
... Group chair is Anssi from Intel.
... One function requests a session between the controlling page and presenting page.
... We also have a function to let the controlling page know about the existence of secondary displays.
... It says whether something is available but not how many.
... If the controlling page requests a YouTube application and you have a TV with a YouTube application, DIAL can be used but some sort of URL mapping is needed for the application to launch.
... This is in discussion within the group.
... The focus initially is just on web content.
... Out of scope things include lower level APIs to discover and communicate between devices.

[Louay is showing a preview of the API code]

Louay_Bassbouss: requestSession starts a connection/session.
... onpresent is a listener on the presenting page.
... postMessage and onmessage are used as in web messaging, with the addition of a close() function.

[Louay is showing an example of the API in use]

Louay_Bassbouss: In the working group there's an update on this function - startSession and joinSession
... This is to allow for sessions that have been started and already exist.
... In the presenting page there's a navigator.presentation global object.
... and a navigator.presentation.session object
... You can use this API to launch apps on a TV and to trigger devices to start from a TV.
... For implementations, there are two possibilities - single user agent (e.g. WiDi, Miracast) and two user agent (e.g. Chromecast)
... We have a FAMIUM prototype implementation made by Fraunhofer FOKUS.
... Another implementation is a Cordova plugin.
... There are demos and videos on YouTube.
... So that's the current status. Please let me know if you have questions.

Oliver: You talked about the abstraction of TV technologies that are already available. Do you have any more details?

Louay_Bassbouss: You can use these technologies without any extension but it depends on what the browser supports.
... E.g. Google is planning to support this in Chrome and they'll support Chromecast only at first.
... But maybe e.g. Firefox will support Miracast.
... So you could have the same application in both browsers but the different browsers would have access to different devices.

Oliver: So providers will have to be careful about which browsers can be used.

Louay_Bassbouss: Yes.

tidoust: It's a complex topic and an interoperability issue. We'd like browsers to support different protocols.
... As you mentioned, the working group has just been started. Thank you everyone who supported its creation.
... Please join the WG if you're interested. We're in the process of setting it up. Ask your AC rep to add you.

<tidoust> [To join the Second Screen Presentation Working Group (member-only link): https://www.w3.org/2004/01/pp-impl/74168/join ]

Igarashi: On Wednesday we're having a breakout session about how to use this API with web of things devices.

Louay_Bassbouss: That discussion could also address things such as playing audio on your home system.
... Also sensors is something to consider.

<kaz> Igarashi's breakout session proposal

Louay_Bassbouss: The API can be a good starting point for that.

Mark_Watson: We're interested in using this API for sending content to e.g. a Netflix app.
... We're hopeful this API can integrate with the DIAL protocol.

<Louay> Presentation API Slides http://famium.fokus.fraunhofer.de/webscreens/apps/slides/presentations/presentation-api-tpac2014/index.html

HTML WG Accessibility TF

<JF> www.w3.org/WAI/PF/media-a11y-reqs/

<JF> http://www.w3.org/WAI/PF/HTML/wiki/Media_Accessibility_Checklist

JF: I'm John Foliot, a member of the accessibility initiative at W3C.
... Several years ago we looked at accessiblity requirements for media.
... [http://www.w3.org/TR/media-accessibility-reqs/]
... We have the spec here and a simpler version as a checklist.
... [http://www.w3.org/WAI/PF/HTML/wiki/Media_Accessibility_Checklist]
... The goal is to get this published as a note by the end of this year so if you have comments, please let us know.
... The document is split into two things - creation of accessible content (e.g. captions, transcripts, described audio/video), and system requirements (e.g. tools for playback)
... As we went through all the requirements, we tried to map it to existing W3C accessibility publications.
... E.g. we have WCAG (Web Content Accessibility Guidelines) and UAAG (User Agent Accessibility Guidelines).
... We've tried to provide this as a resource for different user groups.
... When you're creating content, make sure you include e.g. captions

<kaz> WCAG (Web Content Accessibility Guidelines)

JF: And when you create a system to enable playback, make sure you provide the ability to control it e.g. without a mouse.

<kaz> UAAG (User Agent Accessibility Guidelines)

JF: We map to WCAG which has become a legal standard in many countries around the world.
... We've requested feedback. In this room, you might especially be interested in the system requirements information.

JF: It's also relevant for second screen - sign language on a separate device is a use case.
... We're meeting on Thursday in the morning from 9:00am if you'd like to join us.

Mark_Sadecki: What we'd like from this group is that if there's advancement in tests or prototypes, we'd love access to that so we can look into accessibility.
... Please join the meeting on Thursday to discuss it or ask questions.

The media requirements document: www.w3.org/WAI/PF/media-a11y-reqs/

The media accessibility checklist: http://www.w3.org/WAI/PF/HTML/wiki/Media_Accessibility_Checklist

Mark: Is there any testing efforts here?

Mark_Vickers: There's no testing framework being created now and the testing task force has finished.

JF: I'm watching discussions around EME. Ancillary content is important - if the media is encrypted, what does the transcript (or clean audio) like and is it exposed to the relevant APIs?
... There's a concern there and we're trying to investigate it so I'd love to speak to anyone about that.

<Zakim> tidoust, you wanted to invite people interested to join the WG

kaz: The accessibility task force helped us to generate use cases and they'll be discussed here this afternon.

<kaz> second round use cases (including accessibility ones)

yosuke: Meeting adjourned. We'll meet back here at 2pm - now it's time for lunch.

New Contributions from IG members

<alex_deacon> scribenick: alex_deacon

Moderator of this session is Glenn Deen - NBCUniversal.

Glenn_Deen: SMPTE Update - Open Binding of IDs to Audiovisual Essence 24 TB Study Group. How to bind a content ID in a way that survives compression and distribution thru the supply chain. https://www.smpte.org/standards/reports/
... There is technology available to do this. SMPTE will establish a working group to develop a standard.
... The hope is to have a standards based mechanism that will contain an ID for media. (most likely via some watermark technology)

Mark_Vickers: What is EIDR and AdID

<Lingo> http://eidr.org

Glenn_Deen: EIDR is a unique ID associated with content.

Mark_Vickers: Would SMPTE define an API?

Glenn_Deen: SMPTE would not do this, but this is something W3C could do this.

Pascal: How does this relate to the ATSC work discussed earlier.

Glenn_Deen: Some overlap between SMPTE and ATSC.
... GGIE - Glass to Glass Internet Ecosystem

<MarkVickers> www.ad-id.org

Glenn_Deen: Trends behind this idea. Many orgs, consortia and companies involved in the complex digital video ecosystem.
... There is a need for an org to look at the “big-picture” and to help coordinate between orgs/consortia/etc.
... Digital content is reaching a scaling well. more users, more devices, more markets, more bandwidth.
... Profession and non-professional tools and workfolws are intermixing.
... Video delivery is getting much more complex - many many standards required. GGIE is meant to coordinate and facilitate.
... video data has exploded 256x
... Scaling issues: Bandwidth struggles to continue to meet demand, growth of new devices, growth in users.
... GGIE working on use cases: Use Case #1 - Edge device selects best source for digital content delivery.
... Use Case #2 - Edge device offers content to other devices (edge device-device delivery).
... GGIE Goal - Help improve the state of the art in open standards for both professional and non-professional digital video focusing on all phases of the digital video life cycle
... Capture-Edit-Package-Distribute-Find-Watch
... GGIE is not a new SDO - Focus on identifying work in existing SDOs.
... In scope - All content personal, prosumer, professional. whole ecosystem, not just the web. Content workflows.
... Focus on Scalability, content identification, metadata, user identity, privacy.
... Out of scope - non-Internet Delivery. Codecs/Encoding. Encryption algs. Legal Topics. DRM.
... GGIE focus will be on use cases only.
... GGIE Outputs: Use cases and gap analysis.
... who has been talking about GGIE. Comcast-NBCU, Mozilla, Cisco, ISOC, Ericsson, Civolution.
... GGIE needs an organizational home such as W3C.
... Proposal is to start GGIE as a “sub-thread” of Web&TV IG.
... Are we going to big.

Jean-Claude: GGIE is a beautiful idea, but doesn’t see a workable path to achieve all that was discussed.

Glenn_Deen: Initial focus should be on low hanging fruit.

yosuke: we should break this big problem into pieces - to get some early wins.

Glenn_Deen: crazy to do all things at once. need to prioritize and focus on biz and user priorities.
... Things GGIE is a great opportunity - the time is right. Will benefit content creators and consumers.
... there is more UGC uploaded to youtube in 2.5 days than all studios have created in 100 years.

Mark_Vickers: Scope is broad, but now that we have specific goals it is within reach of this IG. Liaisons will be key. As we are limited to requirements the scope is more managable.

Jeff_Jaffe: Thinks IG is a good place for this work.
... Do we have the right stakeholders in this IG? Can we make it sucessful? Do we have a list of stakeholders not currently in W3C to ensure this is effective.

Glenn_Deen: No list exists.

Jeff_Jaffe: Need to look for both tech gaps and ecosystem gaps. Need to think globally.

Glenn_Deen: Challenge is to bring together a diverse community of specialists. Feels we can quickly fill gaps as they are discovered.

Bill: Trying to get the structure and first steps (binding ID via watermark for example) will get the ball rolling.

... Perhaps we need a task force for a next step.

Update on HTML5 EME, MSE and in-band resource issues

<jcverdie> scribe: jcverdie

[MSE and EME by Mark Watson]

Mark: MSE. Candidate Rec in July. Stable since then
... deployed and used quite extensively
... EME, Discussion continues (and will for a while)
... The TAG recently published opinion on EME and raised a number of issues
... Still a few things are being discussed
... Good news is that it's deployed in various desktop browsers and being used in various apps including us (aka Netflix)
... WebCrypto: stand-alone API to give access to cryptography from the web browser
... Ready to exit LC and move to CR
... API is implemented and used in the field in various forms

Raphael: Overlap between media segment and media fragment spec?

Mark: not familiar enough w/ Media Fragment to answer

Raphael: State of bytestream registry in MSE?

Cyril: I don't know

<MarkVickers> https://dvcs.w3.org/hg/html-media/raw-file/tip/media-source/byte-stream-format-registry.html

Mark V gives more details on this

Mark Watson: append doesn't only mean append to the end, it can be "on top of" already placed media
... you can do any kind of slicing you like

Mark Vickers: example of audio track with various languages

<ddorwin> Gapless audio playback with MSE: http://dalecurtis.github.io/llama-demo/index.html

Glenn Eguchi (Adobe): difficulty if different part come from different providers

ddorwin: I'm not an MSE expert. But you should use the mailing list to ask

Mark Vickers: is this because you want a quick 1.0 version or is there another reason?

ddorwin: I'm not sure tracks is well implemented at all

MarkVickers: HTML5 is going to REC tomorrow ... spec features have to be implemented by 2+ implementors to be considered final
... I want to mention that dataCue has been moved from 5.0 to 5.1 because of only Safari implemented it.
... Those of you interested in this critical data stream access,
... maybe you can help to get more support in 5.1 with additional implementations.
... Need more useful content to justify the implementation. Chicken and egg problem

[Daniel Davis on in-band media tracks]

Daniel: Quick update, I'm not active on the Media Resource In-Band Tracks CG
... "How to expose in-band tracks as HTML5 media element video audio and text tracks"
... most of the work focuses on MPEG-2 TS now

(link available from the slides)

<MarkVickers> http://dev.w3.org/html5/html-sourcing-inband-tracks/

Daniel Davis : The CG cannot create a standard but the work can be prepared to pave the way of an actual WG work

scribe: HbbTV consoritum brought a proposal to increase DVB support in addition to ATSC

Mark Vickers: if you have an app consuming a particular piece of media, you would like a inband tracks to be the same and found with the same name across all the browsers

scribe: this is *not* a req to say that a browser should support MPEG2-TS or whatever
... more *if* you do support, please follow this

Yosuke: HbbTV provided the "european" vision for DASH and MPEG

John Folliot: is the list in the wiki a comprehensive list?

Giuseppe: I think these are coming from HTML5

(we are talking about captions, subtitles, descriptions...)

John: it seems there are new ones defined here

Cyril: this list was build from HTML5
... nothing was added

John: there are only 5 values specified in HTML 5

Cyril: the wiki hasn't been updated

François: the spec says the same

Cyril: the current draft has a general introduction. rules for mapping in-band tracks to HTML text, audio or video
... some in-band tracks won't be exposed as they would be too much expensive

<giuseppe> the list comes from the html5 spec http://www.w3.org/TR/html5/single-page.html#dom-audiotrack-kind

Cyril: in some case on track might be exposed in multiple tracks
... There are 17 bugs opened on the spec but you're welcome to submit more

Afternoon break, we'll reconvene at 4pm

<MarkVickers> The different track kinds are in two tables: http://www.w3.org/TR/html5/single-page.html#text-track-kind & http://www.w3.org/TR/html5/single-page.html#dom-audiotrack-kind

The Second Round Use Cases

<yosuke> https://www.w3.org/2011/webtv/wiki/Media_APIs#Submitting_New_Use_Cases_and_New_Ideas_for_Second_Iteration_of_Work_in_2014

<giuseppe> scribe: giuseppe

<scribe> scribenick: giuseppe

[Yosuke provide overview of the approach, as described on the wiki]

Yosuke:first iteration of UC/req/gap was done last year, result was the TV control API CG
... second iteration started in May 2014, should be finalized by end of this year

<yosuke> https://www.w3.org/2011/webtv/wiki/New_Ideas

Yosuke: On the wiki, there is a list of UCs submitted by members
... we originallu had 7 UCs, one was idenftified as already covered
... of the remaining 6, one (triggered interactive overlay) was identified as in scope for the TV control API CG
... the UC about clean audio came from the accessibility TF
... is about being able to separate the main dialog from other audio source to be able to isolate it
... we examined this use case and we found out that the web Audio API is already a good solution to equalize existing audio track to fit individual's hearing character
... so the only gap was about synchronization of multiple audio/video tracks
... so it was folded it in the media synchronization UC.
... first let's look at fingerprinting
... we looked at it from a user agent perspective
... although there are different algorithms there are some commonalities and there should be a way to have a unified API to talk to that module

giuseppe: do we really need an API for watermarking? isn't the track API enough to "extract" an ID associated to a piece of media?
... or was the idea to allow watermarking implemented in JS?

yosuke: We don't stick to create new APIs.
... If that's sufficient, let's go with that. But we still need to define how we map watermaking and text tracks.

@@@: suggestion to avoid a plugin/modules approach, as it makes standardization difficult.

yosuke: I share the point. However, module system should be an architectual option thanks to EME, at least on early stages.

Dongyoung: do we really need a native module? can't this be imeplemented in Javascript?

defining a generic API could be good anyway, someone could also then implement a polyfill for it

glenn: did you consider video fingerprinting?

yosuke: not really, we started with audio as is was the main use case.
... we decided to focus on audio, but just because we had limited time

jc: what is the value of a fingerprint API? ... can't an app just send the audio fragment to a server and get back a result

yosuke: as you need to generate an hash before sending to the server, we thought you need an API to get this hash before sendding to the server

yosuke: is there an overlap with the GGIE work?

Glenn: probably there is a lot of overlap, although we may be looking at slightly different use cases

yosuke: What do you think about dealing with this in the "GGIE" TF?

Glenn: We should go separate with sharing information.

yosuke: Sounds good. ... let's move to mediastream synchronization
... we identified that Media Stream Synchronization is a basic feature needed to cover different use cases
... we want to explore this further to see if this is covered already by under standards

jc: I think UPnP has something about it. Starting from syncronizing system clocks.

yosuke: we can use UPnP as an underlying mechanism but still lacks UA APIs.

giuseppe: Media Controller API provides syncronisation feature for HTML but we may have something we should enhance it with.

<MarkVickers> http://www.w3.org/TR/html5/single-page.html#mediacontroller

giuseppe: also HbbTV and DVB have done work in this area

yosuke: also hybridcast.
... HbbTV and Hybridcast needed to extend Web standards to achieve sync
... between broadcast signal and content from the Internet.
... that implies Web standards miss something to achieve our use cases
... we need to discuss what would need to be defined in W3C then

yosuke: what should be next step?

giuseppe: before we bring it up to a WG need to identify what is missing

yosuke: ok maybe we should look at existing work
... where to do this work? let's continue here in the IG.

Wrap-up & Next Steps

<yosuke> scribe: yosuke

<scribe> scribenick: yfunahas

Giuseppe: Each session had its next steps.
... I just want to walk through the result as the wrap-up.
... We'll create a TF to deal with GGIE topics.

Mark: We raised awareness of TV guys about what's happening in other groups and how to join.
... And Media APIs TF will continue its work: audio fingerprinting and watermarking, and media syncronization.

Daniel: TV Control API CG has decided to move on to the next step.

<jcverdie> tentative breakout schedule https://www.w3.org/wiki/TPAC2014#Session_Grid

Jean-Claude: What about "screen-less" second screen presentation API?
... startSession is scheduled at 1:15pm in Sequoia

Igarashi: We'll talk about it during a break-out; the CG will continue as a public discussion group.

Louay: I'm not sure that WoT view will fit the CG, which is still focus on presentation devices.

Igarashi: I'd like to see what will be the conclusion of the break-out session.

<tidoust> Breakout session on startSession for screen-less APIs: https://www.w3.org/wiki/TPAC2014/SessionIdeas#startSession.28.22WoT_devices.22.29

... Your opinions are welcome.

Giuseppe: AOB?

Igarashi: On Wednesday, we'll talk about not only screen-less devices but also privacy and security of devices in home networks.
... HTTPS is a major way to achieve Web APIs security but lots of home devices don't implement SSL; This is an issue.

David: CORS is also an important part of security of UA. What about this?

<ddorwin> http://www.w3.org/TR/mixed-content/#authenticated-origin

<ddorwin> http://www.w3.org/TR/mixed-content/#is-origin-authenticated

igarashi: it's also an important issue. i mean how to define origin in home network need some work.
... let's continue this discussion in the break-out session.

Giuseppe: Thank you for joining the meeting. The meeting is adjourned.

Summary of Action Items

[NEW] ACTION: Bin to update the gap analysis to indicate what is at an OS level. [recorded in http://www.w3.org/2014/10/27-webtv-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-11-07 09:18:41 $