W3C Logo Ubiquitous Web Domain

Minutes of Second W3C Web and TV Workshop / day 2

8-9 February 2011
Hosted by Fraunhofer-FOKUS, Berlin, Germany

Netflix Logo
IPTV Forum Japan Logo
Tomo-Digi Logo

Workshop Support

Open Media Web

FP7 and EU logos The workshop has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement n°248687 (Open Media Web) and n°257103 (webinos).

Check the Sponsorship program and become a sponsor for the Workshop.

Date: 09 Feb 2011
See also: IRC log

This page contains the minutes of the second day of the W3C Web and TV workshop that took place in Berlin on 8-9 February 2011. The minutes of the first day are available in a separate page.


See the workshop agenda for details.


See the list of workshop participants.
Francois Daoust (W3C)
Jean-Pierre Evain (EBU)
Giuseppe Pascale (Opera)
Stephan Steglich (Fraunhofer-FOKUS)
Francois, Christian, Olivier, Chaals

Session 5: Metadata / Semantic Web

Moderator: Jean-Pierre Evain (EBU)

See description of the session in the agenda for links to papers and slides

Web Standards and Rich Media Experiences on CE devices, by Gregg Kellogg (Connected Media Experience)

gregg: [presenting CME, goals, underlying technologies]
... Principle components: a manifest, HTML file containing the description of the content.
... The furniture would be aspects that are not necessarily described but implied by the ontology
... We're consolidating on a W3C widget package.
... [example of a generic media player on a PC]
... Track interactions, immersive type of experience with any type of presentation.
... Same package can be sent to multiple devices (mobile, TV).
... That's not to say that everything is uniform. One of the advantages of a semantic presentation is that we can describe things in a more virtual sense.
... Two profiles so far: a mobile profile and a high definition profile.
... Profiles rather describe different encoding, not displays.
... [example of a CME manifest in notation-3]
... Artists may have contracts with specific constraints on what can get on screen and what cannot.
... Different relationship can be described using the ontology
... Using unique identifiers enables external sources to reference CME-released tracks (example of DBPedia)
... Basic workflow: go to a retailer, buy content (MP3), the media player identifies content that may be CME through some ID3 tag, and performs discovery to enhance the presentation.
... The proof of purchase is extracted from the MP3 file
... [example of a proof of purchase with a signature]
... Multi-platform, presentation is based on HTML/CSS/Javascript
... The agent architecture is an HTML5 platform with a few specifics.
... One of our expectations is DVD-like stability (it needs to run in 2021).
... But there are some limitations such as the online support as a release may not come with support for a long period of time.
... Using widgets that provide persistent storage could be a solution.
... HTML5 is not yet done.

jp: Put in the context of Web and TV. First reaction is let's not forget radio, music.
... Maybe we can consider solutions in these other fields.

jp: The media annotations group has been done some stuff.

gregg: the media annotations is more a flat thing. We need more structured information.
... I already presented this work at last TPAC in the media annotations WG. Daniel Park is contributing to CME as well.

Metadata and Service delivery architecture, by Alexander Adolf (Conditon-ALPHA)

alexander: I've been working a long time in DVB.
... [worldmap of DVB land]
... Countries where DVB is already deployed and countries where there was a signed agreement on terrestrial frequencies.
... Time scale for these decisions: two times per hundred years.
... 500 millions receivers deployed by DVB members as of late 2009.
... TV is very huge.
... On total digital receivers in use: DVG is 52%, IP is 7%. It has momentum.
... Technical proposal: example of twitter feed with a link to the TV show
... Users click on the link, and get the show.
... It would be great, indeed.
... The scheme part is "crid". I'll get to that later on. The contextual menu should have three entries: Read more, watch and record.
... These 3 options call for two features: one is service discovery and selection. The other is content guide.
... Let's focus on the content guide. DVB has used TV-Anytime, based on XML.
... It's inherently Web-friendly.
... It allows for segmentation, and schedule, and uses CRID (Content Reference IDentifier)
... All features (title, director, ...) link to a CRID that identifies the content.
... This name then needs to be resolved: content you have, broadband accessible content or broadcast.
... The URI resolution process is one-step: DNS servers give you the address. CRID resolution process is multi-step.
... example of Olympics and news, with resolution process that take you to things that are more specific.
... We do not have such a resolution process as of today.
... We need a federated and distributed CRID resolution service with a first level local component in the client, and security and authentification.
... Content providers want to ensure that the content displayed is their content.
... Now the other part: service discovery.
... [example of a service discovery bootstrap]
... Two modes of delivery HTTP, where you do GET requests. If you get a DVB address, you do the other thing.
... Service discovery records already use XML.
... So why do we need this complexity?
... Service provider takes content from Content creators. Platform operator takes content from service provider. And the network operator delivers the content.
... Platform operator and network operator may be combined. Some people do everything, or almost everything.
... All combinaisons are possible.
... I could continue this for a long time. That's why it's important that we have fine-grained features.
... Standards proposal: I don't think that DVB will really want to start working on a full internet TV solution. They're a big player, but not to the point of stepping into other businesses.
... I think we will need to team up, between DVB, MPEG, W3C.
... W3C has HTML5 and CSS3. DVB has already created XML formats that could be added. There's a set of 10000 Java APIs that Blu-Ray has also adopted.
... As you all know, DVB has been working a lot with MPEG.
... We need more connection between W3C, DVB and MPEG.
... Interactivity with APIs is required.
... Conclusion: we can have a willing combinaison in implementing this. My goal is to have service compatibility.
... Services transported on different networks.
... I think that would be a winning team.

Question: consumer electronic in the US has something built on TV-Anytime.

thomas: why don't you use HTTP URLs?

alexander: how do you ensure that the metadata still corresponds to the content? Whatever operating environment you may have. I think we need that additional indirection.

jp: there are also DVB URIs.

thomas: Still, HTTP URLs are much more deployed than everything else.

alexander: in my view, the source in the video tag would be a CRID or a DVB URI.

mark: getting more info from a URI about other URIs is very much REST, common API functionality.

giuseppe: what about the rest of the world in the map you showed?
... For the audience that comes from the other parts of the world, do they share your views?

alexander: In mobile, we went from segmented standards to global standards. We're heading in the same direction for TV.

giuseppe: some requirements were identified from some region, are they valid elsewhere?

jp: that's what the IG is for. Join the discussion and IG, to reconciliate the URI resolution mechanism as well.

NTT on Service Discovery, by Kiyoshi Tanaka (NTT)

kiyoshi: IPTV Forum Japan has 54 members and develops specifications followed in Japan.
... I'll show a video originally intended for promotion, but it's a good starting point, I think.
... [video presentation of IPTV Forum Japan]
... Introducing IPTV with the example of Hikari TV.
... Screenshot of the menu on a TV screen. With a remote controller, you can interact with features.
... Two features: high-definition video delivery and IP retransmission of digital terrestrial contenct.
... Specifications are harmonized with digit TV broadcasting service, so we can manufacture TV that are both DTV and IPTV.
... They are also harmonized with ITU-T.
... There are many points in common.
... IPTV terminal is based on H.721.
... This model is implemented and deployed.
... Going back to service discovery. Why is it needed? Multiple service providers available over each network.
... You need some way to discover them when you use a new IPTV device.
... It allows users to enjoy various services and service providers easily. You need to get service provider descriptions for the user to select the appropriate service provider.
... ITU-T H.770 defines the various parts needed for that selection to happen. It is harmonized with DVB, as well as ATIS (Alliance for Telecommunications Industry Solutions).
... In the Web and TV, they must be some mechanism to discover service provider discovery and service discovery, harmonized with ITU-T H.770.

hj: are there providers in Japan that use the same specification?
... All the service providers use the same specification in Japan?

kiyoshi: service providers can use the standard, yes.

hj: could this be used outside of Japan?

kiyoshi: I don't know.

masahito: wearing my ITU-T. Developing countries in particular are adopting ITU-T right now.

<chaals> [India, China, Singapore, Nepal... and many developing countries.]

jp: We have a proposal here to have the IG have a look at the specifications from ITU-T. Can W3C have access to H.770?

masahito: there is an official liaison between W3C and ITU-T.
... Everybody can download the PDF file of these specifications.

jp: ok, we need to ensure that the work flows in the right group

masahito: in Korea, some specifications of ITU-T have been adopted. TV-Anytime is being used there.

APIs and URLs for Social TV, by Dan Brickley (NoTube project & Vrije Universiteit)

danbri: going to present 3 requirements from linked social tv, and will try to reshape the requirements in terms of what W3C should do next.
... [some background on danbri, w3c, semantic web project, joost, social web group]
... We looked at different things that failed to build or were built to extract actual requirements.
... First thing: let metadata flow widely, advertising content rather than be a hidden asset. Second thing: it is very useful to identify content with useful URLs.
... The scenario we have in mind is not TV as a device. The Web is not just technology. It's a linked world with lots of people connected to it.
... We sketched a lot of scenarios.
... and we noticed that nobody says stuff like I wish wacthing TV was more like using a computer!
... We build some prototypes and demos. I love APIs too.
... Here is an example: you can use the Web to put stuff on top of others. Sometimes it's useful, sometimes it's not.
... I've brought a few remote controls [showing a huge TV remote control]
... People who designed these things were not stupid, they had to resolve competing requirements.
... We tend to think that by hacking with free software and media center, we can experiment freely.
... We've made experiments with pairing protocols, for instance with QR codes and XMPP uris.
... [demos of prototypes with second-screen]
... We wanted to build things that are agnostic of the device you're using, the context you're in, etc.
... We need to take into account new devices such as multi-touch big tables.
... Second screen APIs have huge potential, standard protocols are needed. XMPP has some potential, but it's hard to address latency issues when you're going through e.g. Google.
... Things are useless without identifiers.
... Example of a demo that connect archive.org films via wikipedia to the rest of the world.
... Whenever one thing is improved, through links, the rest is improved as well.
... We've used Microsoft Pivot Viewer here.
... By using a URI to identify content, you can follow links and get deeper in the Web.
... If you look at the source of these two pages: you'll see RDFa with Facebook ontology.
... Conclusion: this is not rocket science (we know the Web is built from URLs). Linking documents is much easier than linking devices. You should be talking to each other more.
... Protocol deserves a working group. I would like to see best practices note discussed within the IG.

[Matt Hammond, BBC taking the stage]

matt: we think we should leave the UI up to the client, and expose the identifiers, metadata, state and a few APIs.
... What we've been building is a RESTful web API, based on XML.
... Discovery mechanisms are specified. Fairly high-level data model (content, sources, outputs, acquisitions, application extension mechanism). These kind of model applies not just for TV, but also for Internet radios or media centres.
... We've built a prototype using MythTV with an iPhone client.
... Accessibility is built-in.
... Things get interesting when you combine things with the Web, e.g. a personalized BBC page that provides info on what you're currently watching.
... Companion content in a second screen (synchronized) is a useful use case.
... We want to push this kind of API. Our draft will be publicly available from Friday, and will be submitted as a member submission to W3C

jp: I would be ready to sign the metadata freedom act, danbri ;)

Session 6: HTML5 and TV: Gap Analysis

Moderator: Giuseppe Pascale (Opera)

See description of the session in the agenda for links to papers and slides

DAE Objects and <video> extensions, by Jan Lindquist (Ericsson)

Overview of agenda ...

jan: [slide of OITF reference architecture]
... several extension to access tv related functionality, including widget management, DLNA RUI architecture, DRM, media playback, PVR, parental rating and communication services

(OITF = OIPF compliant terminal)

jan: scheduled content, including service discovery

[Slide with extensions for playback IDL ]

jan: Classes for audio and video components (AVComponent) accessible from stream, allows selection of video / audio through javascript
... Events dispatched when new components available or components are removed
... Can be used for adaptive streaming

[Slide with extensions for buffering]

jan: Events and properties related to bandwidth availability (for adaptive streaming)
... Also events for when periods change (for example when changing channel)

[Slide on DRM (Content service protection) API extensions]

jan: Generic interface, without specifying specific DRM system solution
... High level interface, being notified when there are DRM events, and passing messages to the DRM system
... Jan is not suggesting W3C use these exact API:s, but hope to inspire the W3C in API design

[Slide on device Configuration API]

jan: deviceID property, to uniquely identify devices
... API for storing digest authentication credentials on a per-domain basis

[Slide on general video]

question Michael Lagally/Oracle: on the API slides for adaptive streaming - why expose the details of AS to the application, shouldn't it be contained on a system level?

JL: The intention is for the application get some kind of metrics for the content you provide. "Good to know" - like an icon to indicate that the quality has changed. Can improve user experience

M?: No way to influence the situation

JL: Yes there is a way to influence the buffering strategy through the interfaces

[Back to video slide]

jan: Two different embedded objects, one for video-on-demand, one for broadcast video
... Gap analysis / comparison between these objects and the HTML5 video element
... Pretty much in synch ... fullscreen still missing from HTML5?
... Components for audio and video (AVComponent as discussed earlier) are missing from HTML5, this is an area where W3C standardization would be welcome
... Semantics of what happens to streaming video when video playback is stopped, is not taken care of by HTML5

No mapping for channel changing (bind to current channel), circular buffer, and recording also not covered

[Reaching expectations slide]

jan: Establish performance requirements
... Certification process to ensure user experience
... W3C provides a toolkit, other groups provide a toolkit. Not the realm of W3C to get into underlying protocols like DLNA.

[W3C Collaborations slide]

jan: 1: Align the video tag for both HTML5 and SVG.
... 2: Need to establish boundaries: Who does what?
... 3: New interface for recording and control of play out of time-shifted content

Connected TV (CTV) Standardisation in the UK, by Steve Morris (UK DTG)

[Slide about DTG]

[Slide: DTG Connected TV programme]

SM: New section in DTG D-book, part B: for the hybrid case

[Slide on DTG ecosystem]

SM: phase 1: Terminals are hybrid, phase 2: Terminals will be broadband only (PC or iPad for example)

[Slide on HTML working group]

SM: Key challenges: Make it easy to use, coexistance with existing services, a step-change in user experience (UK users are used to DTV services), must be usable as a base for new specifications
... Based on HbbTV, plus new features which are required for UK
... Biggest new feature which is not from OIPF is improved graphical capabilities
... Advanced CSS3 (transformations and transitions) and HTML5 canvas elements
... Subsets to perform well on existing silicon
... Extensions to the application lifecycle, including coexistance with MHEG and other types of applications

[Slide on challenges and opportunities]

SM: users don't care where content comes from, why should applications force them
... Many benefits for users and services

[Slide on cooperation with other bodies]

SM: Not having to reinvent the wheel is key
... Common conformance and interoperability tests is very important
... Otherwise, ttm is simply slow
... DTG is working with other bodies, hope to see W3C as a partner in this
... Core technology is more than just a spec, also a common way of testing

chaals: Talked a lot about testing "noone reads the spec all the way through". How do we share tests for effectively? How to make sure test cases are handed back to spec implementers? To avoid fragmentation of conformance suites

SM: It's about who should a set of tests. In DTG, it builds on work from many bodies. It comes down to being able to say: "This is what we are using". Not necessarily about who implements tests where, but about defining the boundaries and knowing that you can reuse test cases from for example companies like CEA

chaals: Test cases are really hard work, if test cases were sent back to people who wrote the specs - the spec group can help review the test cases and that they are used in the same way. And help other groups get access to the tests.

SM: That is where everybody wants to end up, but because tests are expensive to make - companies who make them are not always willing to distribute them freely

Giuseppe: Lots of work has been done outside of W3C, but why done outside W3C?

Mark Vickers: Taken a very different approach on this nowadays, previously build on top of MHP. Had everything we needed for TV but doesn't make sense on other devices. Now we have services we want to run on other devices. From that POV the only solution is a single common platform. Gather requirements to see what the missing gaps in HTML to arrive at such a platform. This should happen in W3C. No reason to make TV-specific API:s, should be able to run on ot

Giuseppe: How do you envision this cooperation in practice?

JL: Create an API that can be used as an input, and reviewed by W3C groups, jointly working between the groups to arrive at an API.

Home Networking, by Clarke Stevens (CableLabs)

[Slide on web & North american cable TV]

CS: CableLabs is formed to bring together several cable companies to form common standards

(scribe was disconnected, sorry)

Trying to fill in gaps

CS: Opened bug 11326 in HTML5 bug tracker
... Now working on sample implementations to facilitate acceptance to specification

(This is regarding DLNA API in javascript)

[Slide on CableLabs prototype]

CS: Prototype based on WebKit and open source UPnP stack (CyberGarage), exposed functionality through javascript from WebIDL definitions
... Create a UPnP control point in javascript / HTML. Locate renderers and servers

JL: You have provided a UPnP framework, not services?

CS: Right, our approach has not been very comprehensive, interested in working with other partners to provide a more complete implementation

[Slide with prototype UI]

CS: Select your content, and your renderer, and you can control playback through the UI

[Slide on next steps]

CS: Work with browser vendors to implement prototypes
... CableLabs to provide as much information as possible to help other implementers
... Also interested in generalizing to other home networking protocols, like Bonjour
... Big concern is security, current interfaces do not have a sufficient security model
... Similar issues that apply to geolocation and device API
... "Opt-in" model

[Informal poll slide]

CS: Raise of hands - how many would support an HTML5 JS API for selecting multiple A/V tracks
... Raise of hands - how many would support a W3C JS API for local device discovery?

Both questions seemed about 1/3

CMN: Have you looked at Opera Unite? Allows you to serve media from a local server in the browser


CS: Not familiar with that technology, sounds like it might be related

CMN: About raise-of-hands, how many people would actually *oppose* ?

françois daoust: You keep mentioning a UPnP API based on DLNA, can be done compatible but independent of underlying protocol. Concerns about DLNA being freely available

CS: Not aware of the exact license

dong-young: BONDI includes DLNA API, is your API in alignment with that?

<MattH> [ DLNA certification process/costs: https://certification.dlna.org/test-lab.html ]

CS: Only in the sense that they are both based on DLNA, otherwise not. Open to collaborations with other interested parties to converge specs

MH: I contributed the BONDI API when it was alive. BONDI is kind of dead, moved to WAC, all API:s are contributed to DAAP in W3C. Discussion on CEA-2014-B which has defined a DLNA interface.

CS: Events have already overtaken CEA-2014-B

JonPiesing: Privacy concerns around the DLNA API in JS. How to design such an interface with an opt-in mechanism which doesn't make the user think that the UI is broken

JP: Especially difficult for TV viewers. Might be OK in an environment in subscription situation or similar, but otherwise difficult to design the interface

<Marcin> DAP thread re DLNA: http://lists.w3.org/Archives/Public/public-device-apis/2011Feb/0023.html

MV (Mark Vickers): CEA-2014-B would need to be contributed to the W3C, until then W3C should work on one of its own

<chaals> [/me hears an echo from Jon of what Danbri said. People aren't hoping that watching TV becomes more like using a computer... at least in terms of UX]

MV: The javascript API can be described in such a way that it does not need to bring in any DLNA IP

MV: Up to implementers to decide if it uses Bonjour or DLNA

Mark Watson: Exposing services can mean different things, how to expose them in a sufficiently agnostic way? Not straightforward on how these interfaces should be designed

CS: Gateway devices is one approach

JL: What is really the scope of the IG?

CMN: W3C is a lot of different WGs, meet in different ways, work in different ways. Members need to learn how to work inside them. The IG is neutral enough to do this, but it is not the place where requirements will be written.

Members of IG need to dive into the working groups and do the work

CMN: It is important to take those next steps and etner the working groups, this is where the actual work will get done

MV: What is the group for this (DLNA) kind of discussion?

... The right group and right time, and it needs people to get involved

GP: The point of the IG is to act as a starting point

MH: The DAP is now rechartering, if there is momentum hear today, it would be good to get input from people here and respond to Robins mail. This is a good time.

Tatsuya Igarashi: Is the DAP specific to mobile devices or can apply to TV and PC:s?

<francois> Device APIs and Policy Working Group

MH: Originated from mobile world - now moving focus to other devices
... Somebody has to contribute from
... to the WG

CMN: Opera is a browser manufacturer in the DAP group, in our interest to make sure it works for all use-cases
... And we do think it is worth supporting

TI: No strong opinion on who works on API, what about providing a framework for local applications, web applications group may be relevant

Current DAP is not exclusively for mobile phones, APIs are general enough to be used everywhere including TV

[Break for lunch]

Session 7: Accessibility

Moderator: Kazuyuki Ashimura (W3C)

See description of the session in the agenda for links to papers and slides

Universal Design Process, by Mark Magennis (NCBI)

mark: National Centre for Inclusive Technology.
... I felt like an elephant when I came here. What am I doing here?
... I'm a Web and TV accessibility person.
... There is a need for inclusive Web and TV services. Lots of presentation have mention users. Some have mentioned accessibility.
... I'm particularly interested in disabilities and aging.
... 1 in 6 person in EU has a disability for some broad definition by the EU
... From the UN, 1 in 5 will be over 60 by 2050.
... GUI are complex, and they are also being forced into people that are not Internet users to start with, whether they want it or not.
... [video demo of a smarttalk TV that reads EPG]
... other examples will be e.g. contrast-related.
... Access services require captions. Also audio description. Audio description gives you an oral description of a scene for people who cannot see.
... [video demo of oral description]
... [lots of blood seen on screen!]
... Accessible media player is key as well.
... The goal is to do a 10 foot inclusive user interface, using a remote control.
... There is some regulatory pressure. In the US, the 21st Century Communications and Video Accessibility Act mandate broadcasters to produce audio description for video content.
... In the EC, the commission has plans for a generic e-accessibility legislation.
... A lot of that is pushed by the UN convention on the rights of persons with disabilities.
... You can expect more regulatory pressure.
... On the Web, the big spec is WCAG 2.0.
... It's being adopted by national laws in some countries.
... It's technology independent so it can be applied to TV as well.
... On TV, you have lots of other standards, by ETSI for DVB access service delivery, ITC, IEC (text-to-speech for DTV receivers linked to the first demo I showed), and then the UK DTG D-Book addresses stuff such as remote controls.
... My suggestion here is to take all these existing standards and requirements and bring them together.
... All these organizations need to talk to each other to ensure that we create an inclusive platform.
... I'm here to learn than to tell you anything.

kaz: I'm also involved in voice activities in W3C. There is some standardization work on speech going on.
... Accessibility is in scope of the charter of the newly launched Web and TV IG.
... The TV IG will also work with these groups within W3C and external groups.
... I quite agree about the importance of speech-to-text for TV.
... What about using speech as input in next-generation TV?

mark: that's the one area I'm least familiar with.

philipp: on the learning part, we have a whole division that takes care of accessibility. One thing that this division does is reviewing all W3C specs that come out of other working groups.
... These issues are quite well represented in W3C.

mark: thanks for mentioning it. I was not aware of that.

GUIDE - Adaptive User Interfaces for Accessible Hybrid TV Applications, by Christoph Jung (Fraunhofer-IGD)

christoph: a brief overview of what we do in the GUIDE project. Specific focus on elderly users
... we need projects like GUIDE because of the growing elderly society, and there are a number of impairments that need to be taken into account.
... At the same time, user interface technology, speech recognition, touch interfaces, etc.
... Third aspect is content providers.
... Some major gaps we currently have to face: we perceive a huge lack of awareness and acceptance of accessibility in the industry.
... It is time consuming and costly to design things with user involvement.
... The APIs are there, the guidelines are there but may need to be completed with specific Web and TV requirements
... GUIDE can help because it replaces user-centric with improvements to accessibility guidelines, adaptive user interfaces, and tools to simulate user environment.
... At design time, we want to provide tools that simulate things such as visual impairment, or the time it takes for a user to go from one button to another one.
... The second thing that supports Web developers is the guidelines handbook.
... On first usage, we collect users data. Once we have collected their data and limitations, we can adapt their experience real time.
... Considering Web technology we use in GUIDE: set-top box by Technicolor, Opera browser.
... We want to seamlessly integrate in existing user interfaces.
... We're looking into markup languages to describe the UI. Wai-ARIA is being considered.
... We have a few applications for this concept: video conferencing, home automation, content access...
... We're into standardization. We carefully observe the developments in HbbTV and HTML5 here.
... We're also really active in the VUMS cluster in EU.
... Conclusion: we have to disseminate accessibility in the industry. We need progress on standards.
... We may need to define more precisely the roles and responsibilities of browsers, APIs, UIs, assistive technologies.
... For application, we need to identify and specify future application scenarios to extract requirements.
... I think we can also leverage adaptive accessibility in non-accessible technology.
... Research-wise, we need smarter accessibility, adapted to users. We'd like to automate the simulation process.
... Use of the cloud can be very helpful.

chaals: how many people know how to get accessibility of their products checked?
... [counting: 6 people in the room]
... How many people have an idea of the level of accessibility of their product.
... A couple, I see. That's good.
... We saw in the Tokyo workshop that broadcasters and TV manufacturers do very good stuff on accessibility.
... W3C has a way to review accessibility. It may not go well when disjoint groups of people come with different angle.
... Especially when accessibility comes late in the game and report problems with specs.
... There has been precedent in the W3C with tough situations.
... Please think of accessibility as a requirement at the very beginning.
... Otherwise it will bite you in the end.
... One of the things we need to look about in the Web and TV IG is what we can do in that field.
... the amount of captioning in TV is probably way above what exists on the Web today.
... This is more advertisement to listening to these guys, and that people like them or me will be looking into such considerations.

christoph: we had a meeting on Monday. We discussed how to disseminate accessibility. Regulation is one way.
... But it also has to bring something to the industries.
... If you provide services to elderly people, for instance, you can provide services to more people.

mark: quite often, what comes out from companies is that they don't see a business case, at least no improvement of business case.
... There is a lack of hard data to highlight the cost vs. benefits of going down the accessibility route.

christoph: Shadi mentioned on Monday that there is a WAI page that describes success stories.

<chaals> [a reason for not generating the hard data is that it's quite hard to do the necessary research (it's hard to get data on the business case fo documenting the business case...)]

Comment in the room: User accessibility is going to be key for the future.
... I would like to see multimodal interfaces in the future. I don't understand why it's so difficult to see the business opportunities.

MarkVickers: most of the accessibility features are useful for the whole customer base.
... Simple example is watching a video with the sound down.

christoph: that's a point I raised, accessibility is not limited to people with disabilities.

danbri: agree with the last comment.
... If we could make a working group in the Web and TV area, it would be good to have accessibility as a starting point.
... Working group on APIs for second-screen scenarios with accessibility, is that a good priority for you?

christoph: yes, that's one way to do it. People need to be more involved in W3C.

mark: not familiar with W3C work. Accessibility is horizontal stuff.
... They need to be integrated into the whole thing.

chaals: in the good days, W3C members send people to work on these matters, and take requirements into account. On the bad days, some groups do not even think about it.
... Having people available is difficult. Not everyone in the world is an expert in accessibility, and it's hard to find the right people.
... My expectations if we were to work on remote control API, there is experience around, as it's been done a decade away.

GeorgeWright: accessibility in TV has been incredibly better than on the Web. Accessible is tied to usability. The TV industry has got it somehow right.

<chaals> [agree that in many ways the TV industry has actually done a better job than the web at large - I hope we bring all the best of that into the Web (not just the Web on TV) ]

MattHammond: accessibility targets other usage contexts.

Session 8: Profiling / Testing

Moderator: Yosuke Funahashi (Tomo-Digi)

See description of the session in the agenda for links to papers and slides

yosuke: [introducing the session on profiling/testing]

Profiling, testing, certification, by Narm Gadiraju (Intel)

NG: need a TV profile - a set of HTML5, CSS features to create a compelling TV user experience
... today's specs are strongly focusing on PC, smartphones
... which has specific user input devices (keyboard and mouse, etc) and a specific mindset
... an experience with pop up windows and scrolling ads may be expected, or acceptable, when using a web browser on such devices
... this may not be true in the 10ft TV experience
... for one thing, there is often more than a single user
... hence a UI challenge
... service and content providers also come with UI requirements
... new applications bring new requirements.
... premium content, social interaction
... also the fact that content can be present / used on several devices
... mentions the confusion about the HTML5 term. Is it just HTML or HTML5, CSS, etc
... note there are other standard bodies implementing (or refering to) HTML specs
... now we have established need for a profile, let's talk about testing
... we need a test suite developed to check that the profile is suitable for the intent we have dertemined
... the test development need to happen with the development of the profile, with members contributing tests
... the test suite and features can be debugged during the TV profile development
... ensure maturity of the test suite, then it can be used as a basis for certification
... certification brings benefits to device vendors and other interested parties
... device vendors can advertise the enhanced value of their offering
... as a user, too, I can be more confident knowing that a device has been certified
... and I can confidently assume that the web applications of my choice will run on the device
... software developers too know they can develop for the platform
... finally, service providers would be able to develop services for several TV platforms and app stores
... certification does comes with challenges. cost and complexity of developing a compliance program are not trivial
... additional overhead
... my recommendation would be to keep the certification simple.
... e.g. through self-certification
... the tests could be made available to all, device vendors could then use the test suite for devices under test
... and submit test results for approval
... [end presentation]

[Q&A begins]

chaals: how do you know that web applications will actually follow the specs, in the context of the consumer being "confident that devices will run applications of their choice"?

NG: applications may not come directly from the TV, thinking more they could come from an app store or service provider
... (hence adding some form of control)
... but no guarantee

chaals: [questioning whether consumers actually care about certification]
... (anecdotal) evidence that developers don't get it right - the web is not made of stuff you could certify so dealing with stuff that is correctly done is not necessarily of any practical value
... mismatch with a certified/certifiable platform

<chaals> [And after Apple came along and ignored all certification but insisted everything went through *their* certification, Android came and built a market that doesn't bother testing. And outsold Apple as a platform]

<chaals> [Not questioning the value of test suites, BTW - agree that they help us all]

NG: question the assumption that the platform itself is not certifiable

[scribe missed part of answer]

James: there is some value for developer to know that the plaform may be certified

Question: is there some research at intel in the area of profiling or tests?

NG: our experience is mostly through our work in upnp and dnla

NG: UPnP has similar self-certification

Philipp Hoschka: say more about how you know self certification works?

NG: experience with UPnP. Don't have the numbers here but a lot of upnp devices certified
... can dig

PH: thank you, would be useful

<chaals> [I *think* that Narm was questioning my assertion that the vast majority of web applications would not pass certification, but I could have misinterpreted]

<chaals> [I should have been more explicit. The impliaction is that it may be more important to support market-leading apps like google, farmville, facebook (and mid-level stuff like banks) than it is to support correctly-written content - if it has little market appeal]

Stable profile in retail TV products, by Jon Piesing (Philips)

JP: "retail TV products and the need for a stable profile of web technologies"
... people need to understand the range of business models
... vertical pay TV, horizontal pay TV, diagonal models

JP: some subscription based, other not. many variations
... all valid, all have different requirements

JP: will focus on horizontal model
... specificity is that there is no network maker in control
... and manufacturers get income when consumers buy device
... and the consumer owns the device
... with no contract between consumer and service providers

JP: it has a major impact on software updates
... updates have a cost - payment to suppliers, integration, testing, and distribution
... typically, this means that there will only be small amount of software updates, critical ones
... this is not unique to TV
... gives the example of Android tablets system updates
... not a lot of updates coming once you buy a device
... this means we need very stable specs
... because if the spec changes, today's sold device is tomorow's legacy problem
... examples of such issues and hard choices: HTML5, Widgets, UK DTG CTV specs - CSS3, Web notifications
... there's no right answer
... need to stress key aspects
... test pages, streams and files
... downstream specs making their own test suites would be a waste of effort, and a problem for implementors
... most relevant for HTML5 (areas which differ from HTML4), CSS2D and 3D
... question of minimal performance whenever working with graphics
... interesting to explore integration of existing W3C tests into future framework
... suggestions
... organisations will be making (stable) selections from newer W3C specs for TV and related markets
... should explore cooperation on test materials
... note a lot of work on API already done by OIPF

[talk ends]

Open Test bed for Web and TV, by Martin Hahn (European Commission)

MH: we are thinking about whether it makes sense to fund open test bed for webtv
... asking the audience

[no feedback from the audience]

JL: So if someone takes e.g. HTML5 video, from a given working draft, that means that W3C should be careful about changing that bit afterwards.

[question was about dependency and cooperation between standard bodies]

JP: standard bodies have been cooperating and acting on issues in spec references for a long time. Just talk together

NG: mentions work done at DLNA, members of which will be bringing it to the table at some point

chaals: liaison between orgs happens through mail/conversation, not documents
... good question on whether OIPF docs are worth bringing to W3C. should be worthwhile and not too painful

Giuseppe Pascale: testing is as important as spec, yes; not sure it is role of W3C to take care of certification

chaals: would be happier if EU funded several test efforts, than one big official one

MH: european projects are only ever started based on needs from constituency

NG: notes that MH didn't leave the audience a lot of time to respond. too early to decide whether there is or not a demand for test framework at EU level

Jeff Jaffe, W3C: agree test is important - and hard. Let's get help where we can find it. How to use that help is the right question, not just a straw poll

Mark Vickers: the first thing would probably be to have W3C scope that work
... then decide whether to have centralised effort, distributed, etc
... we first need to know what the effort will be

Clarke Stevens: what about other industries, like PC industry, are they pushing for certification

Jeff: want to offer clarification on html5 testing effort
... no doubt that there is a need for robust test suite
... announced work at meeting in november
... starting to get test cases from the community, with browser vendors already contributing

<chaals> [There is a push for more and better tests... but not for certification]

<chaals> [test suites are living and dynamic so can update in the rhythm of web technology development... certification tends to be less flexible]

[Q&A ends]

[short break]

Conclusion: Wrap-up, Next Steps, and Actions Items

See description of the session in the agenda for links to papers and slides

Priorities from TV makers point of view, by HyeonJee Lee (LG Electronics)

hj: [summarizes differences between pc and tv
... The various standard bodies provide specifications and extensions for TV.
... Meanwhile, smart phones, new devices are getting on TV.
... So now, we are discussing the Web and TV in W3C.
... we'd like to make the discussion profitable for our customers.
... From the user's point of view, applications seem to be almost the same.
... But content providers and TV makers need to support multiple solutions. That's a huge waste of time.
... Also, TV is a constrained device.
... So, TV makers want to have a single solution.
... The W3C is the safest place to have baseline platform because the patent policy is clear and RF.
... We think the success of this activity is not related to the technical expertise. The most important criterion is the time to market.
... First thing we should try to do find a common baseline.
... First priority: video tag extensions for HTTP adaptive streaming and DRM.
... If we are successful to have the baseline of that technology, then the integration will be beneficial to everyone.
... To meet our schedule, I'd like to propose a task force within Web and TV IG.
... Second priority: multi-screen interaction with new Javascript APIs.
... Task force in the IG or a WG.
... Third priority: TV profiling. That's a controversal issue. I think we should have a task force within Web and TV IG.
... Possibly give some initial input by end of March.
... Fourth priority: security considerations of Web technologies.
... TV is the most stable device at home.
... malicious code could exploit security holes.
... Fifth priority: Make a good developer guide for content providers.
... 16:9 expected ratio, navigation rules of TV remote controllers, etc.
... To be done later on in the IG.
... We hope that this can all be done in a 1 to 2 year time frame.
... If we lose this timeframe, we cannot prevent chaos.
... As TV makers, we will actively commit to these activities.

yosuke: comment on task forces within the Web and TV IG
... would like to hear from people which kind of TFs should be included in the IG

GP: Just to be clear, the presentation was to provoke discussion ...

JW: Would this be used to provide broadcast stuff, or just back end stuff.

Mark: Will be used for mainstream - all content

Summary of workshop discussions, by François Daoust (W3C)

FD: Here we want to assess the level of interest about different topics.

[FD explains the differences between working and interest groups, and what they can do]

FD: Important basic requirement is people to do the work.
... there are some ideas for task forces within the IG, there may be things we just don't act on, or we could create/recharter working group(s) if there are people to drive the work.
... Not that making a group requires people to be there, and some to drive the work of scoping.creating etc, writing the charter.

HTTP Adaptive Streaming

FD: IG can check on getting RF commitments - making a working group would just lose time.
... intergrating in HTML could require a new WG.

CMN: Why create a new Working Group to integrate stuff in HTML, given there is an existing one?

FD: If we can do it without touching the HTML5 spec that's better because that group has a clear roadmap.

MW: Regarding royalty free DASH it needs to be lear that it is the companies that we have to ask, not the SDOs. It would be valuable to send something from this meeting, to the 3GPP meeting next week.

FD: Any support

CMN: Yes. Support that it is a good idea.

[Moves that this meeting sends a formal request to 3GPP meeting requesting that IPR holders follow Microsoft in providing a royalty-free license to essential IPR]

Danbri seconded.

[Strong support, about 30 people, a couple of votes against]

Alexander Adolf: Warning that it is unlikely to be a resounding yes, since it is in the market already...
... DVB members will have questions back, so the discussion will go on.
... what is exactly the message to go there, what shape does the ongoing discussion take. Please send me a message I can broadcast to the DVB membership to help bring the fence down.

FD / GP: Message will be outcome of this meeting.

AA: What are your formal cooperation methods?

[Action FD to follow up with the request on how to do liaison]

JJ: Appreciate the metaphor of bringing down a wall in Berlin ;)
... suggest that when we write up the request that we amend the motion to be clear that it is for the web use case.

MV: Please track liaisons we should follow - DLNA as well as DVB

GP: Good that people want a clear view of how the proces works.

FD: Official statement has to come from the IG. This is a workshop minus a couple of participants.

MarkW: We should be clear that we don't expect to get back a statement from everybody. There will be a judgement call to make at some time. I think it is important to be precise. So it is useful to identify the W3C-side contacts who can engage in a discussion of what this means.

Iraj: For MPEG, if the IG asks for this it will be more effective that the workshop answering.

Support for additional functionalities

GeorgeWright: We need synchronisation of media.

FD: In what form?

GW: Not sure, but it needs to be listed as an explicit requirement.

JCD: Sometimes the best solution for something wouldn't necessarily be an API, but some other support for...

FD: So the workshop can make a request, and the IG can make its own request later - please join the IG, because it relies on its members, and bring this to the IG.
... Not sure what we need to work on in content protrection. Common Encryption doesn't seem to fit. SUpport for key exchange can be done, seems to be a valuable item.

Olivier Thereaux (BBC): Missing a liaison for the capacity to play content-protected video, extension to canplaytype().

<danbri> [ wonder if everyone here knows how to join the IG ... http://www.w3.org/TV/#tvig needs a bit of update ... ]

<kaz_> kaz: people can bring these ideas to the IG mailing list

?5: It is good to look at what people who are already using stuff from browser world actually do, when we are thinking about profiling.

FD: Not sure what next step in profiling is - think it is discussion in IG do figure out what we should be doing, reviewing what we already know, ...
... related, testing framework.

<danbri> [ http://www.w3.org/2010/12/webrtc-charter.html "API functions for establishing direct peer-to-peer connections, including firewall/NAT traversal " relates to the Remotes / 2nd Screen / "Trick modes" proposal ]

?5: Think they are seperate topics.

scribe: all the acronyms are already using some form of web technoogy in their systems...

<MattH> [ +1 danbri's observation ]

FD: Support for *

Olivier: You are missing things - you have play/pause/ff and there are far more things that people were demoing. Multi-device, multi-screen control and command would be better.

FD: If you generally think it needs to be merged into one exercise that makes sense.

OT: Re DAP charter, it may be currently too limited to a single device - we are talking perhaps about DAP, RealTime and maybe something else.

<kaz_> [ danbri, "how to join the group" is explained in the charter (http://www.w3.org/2010/09/webTVIGcharter.html), but yes, we can put it on the group page as well ]

FD: DAP charter is currently under discussion. Since TV is a device in a broader scope it seems relevant and timely to incorporate these requirements in the new charter.

Tatsuya Igarashi (Sony): Why is DAP being rechartered?

FD: First, because their charter is about to expire. And being aware of that we want to enlarge the participation.

DanBri: Think there is a strong case for remote control / second screen work. It isn't classic DAP, we haven't done the requirements gathering yet, and it requires more thinking than we can do in the time here.

Clarke: Trick modes, downloading content. Trying to figure what DAP is doing. Seems like it has been focused on cell-phone related apps.
... wondering if this spread the scope too thin
... if we add it.

<danbri> danbri: and that there is a protocol aspect, nat traversal aspect, plus the specific domain modelling work for doing specific APIs. but first we need to get these devices to have a communications channel; if that involves NAT-traversal, see the RTC draft charter, which includes exactly that.

FD: Don't want to have a really broad scope. If we don't want to see TV and mobile as verticals, we want to have device-agnostic stuff. The WG was created from mobile input, but looking at device-agnostic design.

SC: Strongly support that goal. Wonder if having different objectives for the API makes sense.
... on one hand, play content, do TV things, other objectives pretty far from that.
... if not a lot of commonality it makes sense to split groups perhaps

chaals: comparison of several groups (HTML, DAP, ...)
... I strongly agree, if DAP doesn't do device-agnostic stuff, ...
... this is about what we're trying to do, not how we're trying to do it
... it's important that this be the goal
... it may be we have sufficent for 2 groups
... not TV stuff, Mobile Stuff; it's not This Workshop ; DAP
... more like some DAP, some webapps, some from here might make sense in a new group, other leftovers might be redistributed

[Unanimous support for trick modes, downloading, etc., about 25 people]

[home networking has strong support (about 25 people) but an explicit concern is privacy and security]

CMN: Think that is a clear and important concern.

Jon Piesing: It's easy to find a solution that isn't very good but gets to market quickly, because doing the right thing takes too long.

Danbri: Seems like you are saying we might do something bad, so let's not do anything.

Jon Piesing: This has been addressed a number of times in a number of places without getting answers so far. It may not be soluble.

PH: This is a similar issue to what is being discussed in areas DAP is already working on - privacy, security, etc.

<danbri> [ I suspect we're talking past each other, because the 'proposal' is simply the text "Support for home networking" ]

PH: think the issue is understood (if not the solution)

Jon Piesing: Getting agreement from a user on a smartphone is different from getting agreement on a TV.

FD: Support for multi-track (issue-152)

Jeroen: Multi-track is part of adaptive streaming, if you put adaptive streaming you will have HTML.
... multi-track. It's probably also tied intimately to content protection as a work item.

MarkV: Could be done. In my understanding of the current HTML process there is a window of this month that will allow putting multi-track, but no existing window to allow adaptive streaming. SO think we should be pushing at least multi-track now ...
... adaptive streaming can be done seperately if necessary.

JW: Proposal currently being worked on is about text tracks. Not convinced that auio will be in browsers before adaptive streaming -think not in all browsers, anyway.

MarkW: Proposal posted yesterday in accessibility area has a deadline of 21 Feb before going to the HTML group. That is one item that it seems like it is progressing and could progress quickly.
... there are a couple of others, and a discussion about whether putting a resource manifest in the src attribute could be done quickly and would be beneficial.

Yosuke: We are looking for consensus on the result of this workshop. This is a global workshop, but not representative.
... IG has public mailing list, and results of this workshop will go there.
... At Tokyo workshop we voted on topics raised, and in summary there were high-priority items.
... We should be thinking here in the same way - check results of that workshop and this group, get consensus on the tasks and priority

FD: Yes. This is stuff to take to the IG... there is a possibility to move some things on a fast pace linked to existing standards work.

GP: There is no reality to us asking for things in priority, the real priority will depend on contributions.
... I don't think it is as important to reach an absolute consensus on what is important, we can see from how much work gets done.

FD: On these items, are there people ready to commit time?

[Olivier Thereaux and Danbri volunteer]

MarkW: wondering where secure device ID goes.
... volunteers to do work.

<yosuke> yosuke: Please be a little bit careful about that what we are doing here is just making a consensus with the participants in Berlin workshop.

<danbri> [ re charter drafts, people should circulate them in the IG, or with W3C team, ... or whatever they prefer ? ]

Jeroen: You wanted something like "get my ID"?

MarkW: Yes. Privacy, obviously, is an issue ...

Jeroen: You have a lot of information about the user...

MarkW: It is important to make sure IDs don't track across domains. The question is whether we should not even strat, or should start and include the security provisions as a requirement of the work.

Jeroen: Seems like you might want to turn off a user.

MarkW: I may block access to my service...

?A: This stuff has strong legal implications, and is very complicated compared to e.g. geolocation.

<yosuke> yosuke: Because workshops are inevitably regionally flavored event. So we'd better treat these results as a good input to the discussion in the IG.

MarkW: Those concerns exist whether you use a standard or not, so the fact (agreed) that this group is not competent to solve them all doesn't affect whether it is done with a standardised technology.

FD: Real Time communication group is starting at W3C

MV: THink this has direct overlap with the home network question - and it should therefore be done there.

?F: We have broadcasters... HTTP adaptive streaming doesn't cover all use cases. Support for multi-cast over broadcasting/networks would be important

<yosuke> And I'm slightly anxious about the fact that in this wrap-up session, the definition of the scope of each topic on the screen seems ambiguous. That's another reason we'd better deal with those items in the IG.

FD: Since we didn't discuss it in the workshop, it shouldn't be in the conclusions of the workshop. Please bring it to the IG :)
... And that is general. Nothing will get done if *you* don't do it.

Danbri: should we do stuff in public?

FD: Would encourage you to do that, but if you are uncomfortable feel free to contact team to talk about it, but please feel free to jump on the list in public.
... [ we had very hot discussion on accessibility as well. So probably we should consider that item as well. ] ... Thanks, gotta go now.

Closing Remarks, by Jeff Jaffe (W3C)

JJ: Good to summarise after the thing has ended and people have run to take a plane
... Thanks to everyone who helped and hosted and stuff.

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011/07/27 20:41:36 $