TPAC Plenary

8 Nov 2017



Dave_Raggett, laudrain, jcraig, CyrilV

Dave, yatil, ericP, chaals-o


<dsr> scribenick: dsr

<Ralph> scribenick: dsr

Welcome & Introduction

Jeff welcomes us to the plenary day

It is a great year for TPAC with a record of over 600 people

Jeff thanks our sponsors: Spi Global, Vital Source and Igalia for helping to put the meeting together

Many years ago the plenary day was largely talks, but we changed the format to encourage lots of discussion.

We now have a couple of talks followed by break out sessions.

Today will we have one talk on web standards and another on testing

Jeff describes the break-out scheduling process

we already have a lot of great ideas for sessions on the wiki and others are welcomed. We have rooms and time-slots for some 40 meetings

Jeff presents a slide summarising core innovation on web standards focusing on browsers (the open web platform)

e.g. WebRTC, Web Assembly, Web Performance, Web Payments, WebVR, Web authentication, Service Workers, Web Components, MSE, and work on Extensible Web

Jeff invites PLH to give the first of the two invited talks

<jeffh> link to jeff's slides?

<jeff> Sorry, jeffh for not providing the link to my slides. Here it is: --> https://www.w3.org/2017/Talks/tpac-slides/plenary-jeff/#sponsors

<jeffh> Jeff_Jaffe_W3C: thx! slide 2 is very useful :)

Improving Web Standardization

<Ralph> Improving how we do Web standards [PLH]

We want to improve how we produce specifications, including the technical report page, the editorial workflow and the repository of specifications

We are building on top of GitHub and adding tooling to make this easier

<Ralph> Philippe_Le-Hegaret_W3C: we have over 200 Rec-track documents in progress on GitHub

We are making it easier to track contributors as part of our royalty free patent policy

We’ve added support for automatic publishing with support for Candidate Recommendations on the way

Good practices for GitHub labels is critical to this

We acknowledge that working with GitHub pull requests can be difficult, but it has many benefits

It facilitates discussion and reviews and helps with automatic publishing

It also helps with testing (spoiler for later section of this talk)

We have around 6000 drafts and 300 W3C Recommendations

We’re working on automatic out-dating of previous versions and managing versioning and links

People have different needs in respect to which versions they want to see, e.g. latest editors draft vs latest W3C REC

We’re updating the /TR page to support filtering and different views etc.

The next major redesign will be in 2018

One reminder is the value of wide review. This should happen early, just before CR is often too late.

Publish often, include questions you want answers on in your specifications, use “wide review” in your status of this document section as tools will pick up on that

Don’t forget to schedule horizontal reviews e.g. security, accessibility, internationalisation, web architecture

For the second part of my talk, I will address interoperability and testing

The set of stake holders, and the framework for conducting tests.

Before 2014, testing was a low priority and test suites were not maintained

We now require a much great emphasis on testing

We now have an open source effort at web-platform-tests.org for a cross browser test suite

This is maintained via GitHub and supported by all major browsers

Please use web-platform-tests.org for your browser specifications

This currently covers 180 specifications and some 1.3 million results per product run

It is time consuming and will work your CPU hard!

We want to check that each feature in a spec is implemented in an interoperable way.

Don’t want until Candidate Recommendation to worry about testing!

Suggestion: all changes to a specification should have tests

Link your spec edits with your tests

Thank you!

Improving Interoperability with Web platform tests

<yatil> scribe: yatil


Philip_Jägenstedt_Google: I’m Philip Jägenstedt, I work on the Chromium project at Google.
... Let’s look at the justification for web platform tests. The web platform works on different devices but there is also a lot of inconsistencies. Everyone on the web tried and failed to implement things cross-browsers.
... Some of those issues exist because of different goals of browsers, but most of them are probably unnecessary. Testing has the possibility to improve it for web developers.

<shawn> visual description of "this" displayed = Tweets about things not working

Philip_Jägenstedt_Google: Web platform test is a big directory of tests for each spec. We have a dashboard now for web platform test and it runs the tests daily on four browsers and you can drill down into different features and see what works and what does not. Sometimes it is very hard to know what is supported.
... Want to improve the browsers, with mobile and edge versions. Up to 8 to 10 variations.
... The goal is that you know that something is supported in an hour after it is put in.
... We are about to start to change how the dashboard looks to bring attention to interoperability foremost. Instead of comparing browsers, we want to compare feature to feature.
... We want to create incentives so that the browsers move towards interoperability. You will have the possibility to see the detailed results, but from a standards side of view this makes more sense.
... We had manual tests, where you need to click something. We now have a way using zoom webdriver that allows web platform tests to do those tests automatically. If your test can be defined in web driver, do so. For more complicated tests, write to public-test-infra@w3.org
... Integral part of standards development. Every normative change needs a test. 100+ specs are using it. Started experimenting with those tests in HTML which was a success. WGATWG adopted, then Web Performance, then CSS WG. In additional a lot of smaller specs and groups are joining the effort.
... There are about 200 specs that are adopted by browsers, so we are at roughly 50% adoption. Hope this gets adapted into the workflow.
... Taking testing more seriously makes sense as soon as you have one implementation in the wild, but it is up to the WGs to adopt it.
... How to do it? First step is just to copy the contributor information from Gamepad to your CONTRIBUTING.md. Spec and test changes need to be in lock-step.
... 2. Learn which implementation fail the tests, filing bugs. If the test still fails after a year, we have to see what we can change in the spec.
... Hope this means less time to interoperability.
... Tests are now a part of the integration workflow. web platform tests wit two way sync in the workflow of the implementers, so working with the browser’s source code repositories.
... The changes can be made locally and then upstreamed to web platform tests. Firefox does it for over a year, Chromium has joined the effort recently.
... Chromium contributions up, people contributing more to web platform tests.
... EdgeHTML and WebKit are working on it. It leads to much shorter feedback cycles.
... There will be a breakout session, I hope, and there is more information in the slides, including mailing lists.
... Feel free to talk to me and enjoy your testing. Thank you!

chaals: You say: Don’t write tests until you have an implementation, but it can be a helpful way for people to think about how things should work.

Philip_Jägenstedt_Google: The only point I make is that it is not always the best way to write your tests early. It might be ok for things that are in incubation. There is not ONE optimal perfect order, it depends.

Ivan_Herman_W3C: There may be specification whose implementation goes through some extension or polyfills. Can this be tested as well?

Philip_Jägenstedt_Google: There are web platform tests that use polyfills but as the tests are not using the error handling, you might have varied results. There is nothing formal we have here.

Ivan_Herman_W3C: Do you plan for supporting this?

<azaroth> +1 to Ivan's question

Philip_Jägenstedt_Google: It is not a technically hard question, we might be able to do it.

Ivan_Herman_W3C: Do we want to do it?

Philippe_Le-Hegaret_W3C: We got a lot of things on our plates, let’s see.

Jeff_Jaffe_W3C: What is the procedure when I as an implementer want to take part of the dashboard?

Philip_Jägenstedt_Google: Currently there is not a process that we have currently, but we certainly want to test as much as is useful to test.

@@: You talked about when it makes sense to test. There are lot of shades in-between. There might be a shipping implementation but the spec needs changes. Should the testing be more flexible for the WGs.

Philip_Jägenstedt_Google: We certainly want to be flexible to make it work.
... I don’t think about it on a spec by spec basis but on a feature-by-feature basis.
... We should figure this out, in the future when we use web platform tests by default, we have the tests by definition when we develop specs.

<npd> ++ for prioritizing new session leaders

Breakouts Preparation session

<cwilso> Is the grid getting updated in the wiki?

<tantek> cwilso - wiki should be up to date

[Breakout sessions Grid]

<cwilso> Ah, now it is - thanks Tantek!

<tantek> caching is hard

<tantek> (for like 30+ min 😂)

<gerald> investigating why our cache setup for /wiki is broken

<gerald> turned off caching in the meantime

<tantek> gerald, not just /wiki - dbaron has determined that www.w3.org home page is sending incorrect caching related headers as well

<gerald> gah

<dbaron> Not sure if it's still happening, but was certainly a big problem a few months back.

<gerald> dbaron, do you see any issues with the current headers?

<Yves> https://redbot.org/?uri=https%3A%2F%2Fwww.w3.org%2F

<dbaron> gerald: yes

<dbaron> gerald: the my questionnaires page returns a 304 Not Modified when that is not true

<dbaron> gerald: so I need to shift reload to make a questionnaire I answered yesterday go away

<gerald> dbaron, thanks, will investigate

<dbaron> gerald: and I think I've had similar problems with other wbs stuff

<gerald> ok

Breakouts Report session

<ericP> scribenick: ericP

Andrew_Betts: we talked about advice for finding polyfills
... terminologies [examples]
... makes it hard to understand polyfills
... is it clogging the standards process or helping us adopt new features

Liam_Quin_W3C: improving web adverts
... had 30-40 folks
... looked at whole ecosystem
... advertisers, channel owners, users...
... there's a BG on how web advertising could be improved
... the advertisers can work with the web instead of fighting the web.

Max_Liu_Alibaba: ... native web
... challenges from mobile developers perspective
... how web tech can help
... what is the potential in W3C, e.g. optizimation of DOM, @@1 access
... setting up a CG to gather use cases

Kajimoto: [WoT Plugfest - Kajimotow]
... We provide a wot-compliant devices
... every device has different protocol [listed]
... OTOH, introducing a WOT arch allows [more] devices to talk to each other
... Panasonic controllers worked with P devices
... we showed that a single controller can work with multiple devices

Marcos_Caceres: [respec, bikeshed]
... most of you are using one or the other
... we answered some questions
... if you like js, respec. if python, bikeshed

Philippe_Le-Hegaret_W3C: [/TR improvement]
... we will have some cool stuff
... notes point to spec-prod

gregg: JSON-LD 1.1 Update
... focus has been on allowing LD to interpret more colloquial forms
... how do we version this so 1.0 processors don't misinterpret 1.1 data?

r12a: [i18n]
... shared our requirements-gathering
... one prob we have is layout features for publishing around the world
... not a lot of documentation available in English
... we try to describe how scripts work in terms of layout
... thinking about instead describing the problems instead
... if anyone wants to describe their own script requirements, please let me know

<chaals-o> scribe: chaals-o

<ericP> scribenick: chaals-o

Florian_Rivoal: [reporting on behalf of proposers of the session: spatial navigation]
... using 4-way controller. on e.g. TV you need this. There are other places it helps
... exists in various places, but there are no standards to make it possible for websites to influence how browsers do it
... We will start a group in WICG. Started from CSS, but also involves components, APIs, ...
... We will develop it until we get bits to put into WGs. There are people who have experience

<ericP> [Subtitle Format Support of TextTrack; #texttrack]

Andreas: Discussed formats, model of text track cue and HTML 5. designed to support WebVTT but there are other common formats
... had a session last year, continued that discussion.
... Agreed it would be worthwhile exploring opening a text track cue API to allow more formats to be supported better.
... Next step is WICG, see where we get to.

<ericP> [Permissions & User Consent - Sam Weiler]

Sam_Weiler_W3C: Large crowd in small room.
... Some surprises - models to avoid asking the user too much, ask forgiveness not permission, if it's OK-ish
... model around accountability - can you tell what the site is meant to do.
... talking about a workshop in Q1 - need UX people to make this a success.
... So please help us find them.

Manu_Sporny_DigitalBazaar: [sovereign web]

Manu: what happens with payments and verifiable claims. We have interesting work on identifiers...
... getting more empowerment on the web to own our own data, and control it
... We didn't have enough chairs...
... didn't focus much on payments and claims because they are running. Lot on decentralised identifiers. Complementary to DNS-based identifiers.
... Looked at credential handler API - how do you move the claims and credentials around the Web.
... Talked about how to move that into W3C.
... There is a credentials CG, will continue the work there, figure out how to move through the process.

<ericP> [Getting in the TAG or AB! - Natasha]

Natasha_Rooney: [TAG / AB]

Natasha: talked about what they are and do, what you should bring to the table.
... not everyone needs to be a genius to join, they are teams and benefit from broad experience across the team.
... TAG elections coming up - please consider running, getting a broader range of candidates.
... TAG/AB learned from the session, what happens when you are running. Will discuss.

<ericP> [Server Proxy Cache Policy - taoqingqian01@baidu.com; #spcp ]

QingqianTao: talked about server proxy problems (AMP / MIP / friends)
... When the page is cached can you send stuff back to the original source when served. Can it be solved in the browser?
... (same origin issues are big there)
... Also, getting the original URL be showed in the browser instead of the server proxy.
... Also, being able to use transition effects between pages.

Vincent_Scheib: [device APIs]

<ericP> [Device APIs: Bluetooth, USB, NFC, Sensors - Vincent Scheib]

Vincent: we have a device WG, looked at things in the space and the charter. We are rechartering, feedback is important...
... primarily discussed recent APIs - bluetooth and USB especially
... reviewed use cases, security and privacy questions.
... see the minutes. Generic sensors and fingerprinting came up.

Vincent: Anssi called for whether there is support on Bluetooth which has developer support

<ericP> [❤️ W3C Stories - Michael Smith]

Mike_Smith_W3C: Only session with a heart, because it is all about love (aka William Loughborough)
... We need to remember the people whose problems we are trying to solve when we do our work.
... remember the stories of what we have done, or tried and what stories do we want to be able to tell in 2020 about how we made things better.
... Shared some stories, talked about where they meet W3C work. Long-term plan is open this up to the world.

Dom_Hazael-Massieux_W3C: [MDN+W3C session]
... Moving MDN to a shared group, with W3C, Google, Microsoft, Samsung...
... how can we synch MDN with W3C better, bring documenting technology inline with defining it.
... E.g. using explainers, looking at compatibility data and testing work.
... Identified some lines for work.

<ericP> [WebVR Testing - Nell Waliczek]

Nell: Future of WebVR API shape. Looked back at the gaps. We need some guidance about how to test,
... want to plant seed about problems developers have in testing. There is a lot of hardware, and buying it all is uneconomic.
... How do we validate implementations, and can developers use that approach?
... Not all testing can be automated.
... Automation can help libraries. Looking at what we can do with WebDriver to control VR
... Action items are to go work on this.

<ericP> [Decentralized Social Network/Games - Chris Webber]

Chris_Webber: Vision is decentralised participatory space, and standards are helping to make this more real
... Showed Activity Pub on Mastodon, sending messages across networks.
... So it works.

Sandro: We have w3c.social here too

Chris: How do you migrate in the cloud. Talked about a "simple" solution - you can upload stuff and not lose everything when the source server died
... User experience is tricky.

<npd> https://joinmastodon.org/ and https://w3c.social/about

Chris: But games? Can we do more. We had more sense of place in the 90s, can we do that again?

<ericP> [QUIC Stats API - Arnaud Braud]

Arnaud: Quic discussion on use cases. We had one - ours, and we could solve it using existing APIs so we can go home early.

<ericP> [HTTPS in Local Network - Tomoyuki Shimizu]

Tomoyuki: Discussed possibility of making local network use secure transport.
... have a community group. Examples, letsencrypt and webauth inspired.
... We are at early stages, need a lot more help.

<ericP> [Credibility "Fake News" on the Web - Sandro Hawke]

Sandro: How do you tell if a web page is legitimate. Where is my Web of trust (and flying car)?
... fake news is a horrid problem people care about.
... There is a credibility coalition

<ericP> https://misinfocon.com/?gi=1f18c3d9847

Sandro: Had a presentation from them about what they use as indicators for credibility.
... What can we do with that?
... We made a CG to do that.
... Dan Brickley talked about Schema.org's claim Review stuff. Search can provide fact-checking statements about things you look for.
... Talked about argumentation and threat models. CG will meet Friday morning 10-11.30

<ericP> [Sonar, linting tool for the Web - Maris Catalin]

Maris: [SonarWhal]
... I have a lot of problems with my name. I am not allowed to put it into info systems :(
... Why we started sonar at Microsoft. Because Web development is hard. fast-moving practices to try and follow.
... went through how to use it, what it does - suggest errors, find old code no longer needed, ...
... Discussed different approaches to this sort of problem.

<ericP> [Browser Extensions - Florian Rivoal, Vivliostyle]

Florian: Extensions started working in one ecosystem per browser.
... they are converging on implementation, nicely. Started a CG a while ago to standardise - specs and tests.
... It fizzled after a while, the meeting today confirmed there is still intent to do it, looked at priorities like willingness to break things to move forward.
... We are still on the same page, so hoping that the group will get more momentum again.
... If you want to influence the work mode, now is a good time to speak up.

<ericP> [3DS, Web Auth and Payment Request - Manash Bhattacharjee, Mastercard]

Manash: Discussion on fraud rates, and insurance issues.
... Countries like India, Singapore, Malaysia, etc are mandating bad experiences to try and stop fraud.
... New concept that could be embraced to improve the user experience.

<ericP> [Spatial Data on the Web IG - Linda van den Brink]

Linda: Need to work on awareness. We are W3C and Open Geospatial Consortium group.
... We had a WG, produced Recommendations on Semantic Sensor Network, Time vocabulary.
... Will now work on statistical data practice, moving objects, looking for other areas where OGC and W3C should work together.

<ericP> [Offlining Documents - Benjamin Young]

Ben: Centred around what we are doing in publishing - polyfill options, seeing the gaps between hope and expectation.
... Discussion about how we might do it now, copying web apps. In the future publishers can stop being application developers and just create documents they can distribute.
... Looked at how to bridge the gap between ebook readers and web browsers

<ericP> [Communications Accessibility - Baoping]

Baoping: looking at accessibility in things like WebRTC, combining with AI to provide visual / sound recognition
... Also looking at 5G and what that offers.
... Goal is to implement services for home users.

<ericP> [Art & Culture on the Web (Museums) - Angel Li, W3C]

Angel: Art and Culture sells better than museums.
... We want to share our cultural heritage across the world.
... Museums are transforming, moving to the Web.
... W3C has a lot of relevant work to help do this.
... And improve the user experience and developer experience. Should continue to explore.
... Did some gap analysis and identified communities we would like to involve.
... Make the Web better for artwork.

<ericP> [Media and Entertainment IG - Chris Needham, BBC]

Chris: Continue discussion in media IG meeting.
... Group as home for audio/video technology work. Putting media content into books, looking at video and VR
... Want to map capability API to media formats
... Talked about video codecs - significant new piece of work is to revise Media Source Extensions with new features - ads, captioning, multi-track...
... in-band events in HTML, and packaging web content inside MPEG containers.
... And how to help Working Groups coordinate with TV-specific bodies when their work touches on TV-related stuff

<ericP> [Practices and Tooling for Web Data - Dave Raggett, W3C]

Dave: Data is growing, e.g. from sensors making lots of it. You can use it to annotate content.
... There are data-sets out there you can download (yay Open Science and Humanities research)
... Looked at Open Data Institute trying to make W3C a better place to develop core data standards that are still needed.
... Please fill out our questionnaire.
... Is the Rec track a good fit for data standards?

<ericP> [Certificate based digital signature using Web Cryptography API - Hyungwook Lee]

Sangwhan: Certificates are used to sign transactions in South Korea using ActiveX. How can we kill it with fire?

<ericP> [Technical Demos - Dominique Hazael-Massieux, W3C]

Dom: Had 10 demos of cool stuff, lots of discussions, fewer people than we hoped. Why? Was the session interesting if you were there?

[I was not there because I was at other sessions :S ]

Thank youse all

[Note the AG WG had a lot of sessions to close more issues in WCAG 2.1]

Thanks to leaders, and scribes.

Coralie_Mercier_W3C: There are 75 minutes until the reception.
... Keep your red tickets to get a drink

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.152 (CVS log)
$Date: 2017/11/13 22:12:13 $