– DRAFT –
TPAC2018
24 October 2018
<koalie>Slides: Welcome & Intro
Jeff Jaffe introduces the session and notes that we have a record number of attendees
Today is for unstructured geeking following a short plenary
We’re going to hear from the TAG (W3C’s technical architecture group) on topical issues
We thank our sponsor
We encourage you all to give us feedback on the meeting
Jeff notes W3C’s code of conduct - let’s be on best behavior
At 9:30 we will have the breakout sessions building
Jeff_Jaffe: I am fired up with the enthusiasm you all have.
Jeff shows a slide listing areas of innovation for the Web
and a second slide on an impressive pipeline of further innovations
Jeff_Jaffe: This year we have a new way of measuring productivity …
[Jeff shows a slide of GitHub commits across W3C]
Jeff_Jaffe: something between a 1000 and 1500 commits a month
... You can see increased activity as groups approach deadlines
... Next year’s TPAC will be in Fukouka in southern Japan from 16-20 September 2019
Questions?
[no]
Jeff introduces the TAG
Dan Appelquist introduces the TAG and explains its role
<koalie> Slides: TAG Update
He explains how to open an issue with the TAG on GitHub
Each issue is assigned to a member of the TAG to manage
more info at https://tag.w3.org and https://github.com/w3ctag/meetings
Dan_Appelquist:We do regular developer
outreach, locally where we are holding face to face meetings
... We have an election coming up and three seats up for grabs
... Please encourage your W3C Advisory Committee Representative to vote in these elections
... We recognise we have a diversity problem, so we encourage you to nominate a more diverse range of people
... It is important that the TAG represents the diversity of the Web as a whole
<koalie>Slides: Security & Privacy Questionnaire (PPTX)
Lukasz introduces the security and privacy questionaire
Lukasz_Olejnik: I am a new member of the
TAG and a security and privacy researcher
... and an advisor to the Red Cross
... It can be difficult to see how a web API feature can be abused to spy on your privacy
... e.g. the battery API and CSS max-blend mode
... Sometimes precision matters, e.g. incognito detection with the Web payment API
... We’ve found the perfect solution - more paperwork. :-)
Short video from Monty Python’s the Holy Grail
Lukasz_Olejnik: We need more clarity and
context during reviews of security and privacy aspects of new APIs
... The new process: privacy by design
... Please seek review early
... Best recommendations in 5 seconds
... you could drop the feature, or sanitise the data handled, and make a privacy assessment.
... You need to consider regulatory aspects - and whether it is a good idea
... We don’t want the security and privacy questionnaire to be a tick box exercise, we want you to think carefully before answering
... We hope to finalise the questionnaire next year
Dan opens the floor to questions
Question: can you tell us what the time commitment is for people on the TAG?
Dan: a weekly one hour call and 4 face to face meetings
... we don’t generally meet during TPAC as we use the time to interact with other groups
... It’s about 20% of your working time roughly speaking
... We use WebRTC for our calls, so we can see each other
David_Singer_(Apple): please tell us about the tensions in best practices and trends
Answer: one example is better
fidelity for Web APIs and a focus on WebIDL and new JavaScript
features e.g. classes and promises and encouraging their use
... These days we’re looking at streams and the best ways
to deal with them
... We’re working up to a thorough review of HTML and
revising features from the ancient past
... We’re trying to explain how the higher level of the
web platform works underneath
... We’re seeing a lot of ideas coming from Google, which
is great, but we want to see more from a diverse range of people
Dan: we want to better support people who find it hard to attend face to face meertings
Phil_Archer_(GS1): is it time to update the TAG’s architecture of the Web which came out in 2004 and is looking rather dusty
Dan: we considered that and decided
instead to focus on the design principles dociument instead
... The Web is more agile these days and we can address things in streams rather than a big tome
Ian_Jacobs_(W3C): web page sniffing is used in payments for risk assessments, how should bona fide needs be addressed
Hadley_Beeman: we have no power and can only offer advice. We encourage designers to expose their thinking including what they’re seeking to achieve, along with an awareness of the potential for abuse
Alex_Russell: we have a very very rich
platform (Dan shows the Unsanctioned Web Tracking document)
... We want to can get sure people make good advice and for browser innovation to improve privacy
Question: can the TAG comment on how well we’re doing in respect to privacy etc.
Dan: it is quite clear that the Web
has had big impact on society, not all good
... We have a tussle between browsers that disable
features for improved privacy and companies that rely on those
features for good business reasons
... We as a community need to come together to discuss
and reach a broad consenus
... You can steal something, e.g. someone’s bag, but
there are consequences under the law. This is also true for the Web
and we need a balance and to ensure that overall the Web is positive
for society
... The Web has a great deal of functionality but isn’t
exposed to web developer, the question is how to safely expose them to
bring parity with native apps
Question: is there much work on consulting end users in respect to API and feature design?
Alex: we are indeed interested in encouraging this, and one approach includes a focus on early review of proposed ideas and showing that they really solve end users real needs
<tantek> "more science and less guessing"
Yves_Lafon: We get feedback by attending
conferences and giving talks
... We encourage developers to write down why they did what they did as this is often missing from specifications
... The lengthy discussion in WGs will be forgotten unless the record is made
(applause)
<nigel> +1
Yves: We tend to review specs twice and find it really useful to have an account of the changes
Question: in our group we like the idea of a technical spec and an explainer spec
... is this a pattern worth exploring more
... (some technical details about role of URL handlers)
[Some discussion about best practice and the need for further discussion]
Question: what about domain specific
languages such as math, music chemistry etc.
... Do we just convert everything to SVG or can we preserve the semantic info?
Dan: Web Components are a useful building block
Alex: we’ve just issued guidance on how to use them effectively
... it would be a shame to lose the domain semantics
... Previously you had to wait on browers to integrate support, but now with Web Components you can do this yourselves
Dan thanks everyone
Coralie Mercier explains …
The sessions start at 11am and continue after lunch. We’ve prepared a grid for time slots and meeting rooms
Session proposers write down their idea on a sticky note and put it on the matrix board
We can then move meetings around to reduce conflicts
Lunch is level -2. Breaks are between 3 and 4 one level up
Coralie introduces the session. Epxlains the procedure. Those who held sessions are invited to come to the stage, and those who haven't done so yet are invited to link sessions minutes or summaries to the Breakout Sessions Grid.
(People are gathering on stage.)
Reports should be 1-2 minutes, not more.
AnnevK: contributing to HTML session.
… Concl is you need 2 implementers on board,
… Contribution need to be standards-quality.
Philip: Web Platform test. MS has built an online test tool.
Anssi: Machine Learning for the Web: Kick off incubation
… High interest. All major browsers there. 60 people in the room.
… Proof of concept in Chromium showed that it can be done.
… Findings from this impl. are that performance is good, close to native.
… Optimizations possible.
… Can be a polyfill on top of WebGL.
… Leaning to exposing both high and low level API.
… First ftf on Friday pm.
Peter: Maps for HTML: Value of lowering barriers to spatial info on the Web.
… Gave a demo.
… Talked about collab between geo-spatial and W3C communities.
… Polyfills.
… Progressive enhancements of those.
… SVG mapping community collab started at this TPAC.
Dan: Web and networks:
… followup on a workshop 5 months ago.
… Wanted to bounce off ideas.
… Exposing hints between appls and networks, optimize without impacting privacy.
… We stated some learned lessons.
… Figure out the principles.
… Mutually beneficial.
… Cheat-proof
… Emphasis on hints rather than strong controls.
… Exploring interest.
… Planning an IG.
Dom: MDN:
… We had people form project advisory board of MDN.
… Discussed results of collab between the various partners.
… Discussed input from participants in the breakout.
… I18n and a11y to be added to MDN
… Workflow of documentation.
… Approach me if you want to know more.
Manu: Decentralized IDs:
… Showed presentation, use cases, next steps.
… Workshop at MS in December
… Proposing a WG early next year.
Joe_Andrieu: Verifiable credentials.
… What other standards do we need to create.
… How to improve our messaging.
… Does Blockchain solve our pbs.
Cyril: Make a community at W3C or join an existing one.
@@@@: [Sorry, scribe doesn't understand what the topic is]
… Proposal is a new feature to set a header to allow browser to create new toplevel browing context.
… We figured out a way for level-2 protection.
… So that web site cannot load resources from another.
fantasai: Spec-editing best practices:
… 1) define the overall model.
… 2) Start with an overview, high-level
… Both overall for spec and per section.
… Example from CSS is how borders interact with tables.
… 3) Keep examples close to the definitions.
… 4) Use illustrations and diagrams generously, but not as a replacement for normative prose.
… 5) Possible modes are constraint-based and @1-based spec writing.
… 6) Esptablish convention for writign about things.
… 7) Read the spec form top to bottom to see if your local edits still make sense in the whole.
… 8) Review the table of contents regularly, see if it makes sense.
… 9) section titles should be evocative.
… 10) Generous with sub-headings, so you can point to bits of the spec.
… 11) Use preprocessor, source control, link checker, and respond to all comments
… Invite the commenter to respond to your answer.
Toshiya:
… I shared my program.
… We discussed how to send haptic info and vision sinultaneaously
… Discussed methods.
… Still research.
… Not ready for commercial use.
… Media IG
Charles: Personalization:
… Goal is to adapt content much more than is currently supported.
… Some people need content without numbers, e.g.,
… Suggestion for attribute "purpose"
… Could also be helpful for others such as Publishing WG.
… There may be other options, such as Web Components.
… Next steps are to explore them.
… We will ask the TAG for a review of our explainer.
Mark: Improving spoken presentations:
… IMS global learning consortium.
… Education space, students facing mispronounced content.
… Content eventually ends up in HTML, so we need support for HTML.
… We have SSML, we had CSS speech, which went mute.
… We will take disussion to task force.
… Will there be voice fonts, like Web fonts?
JihyeHong: Spatial navigation
… Gathered some feedback
… We currently consider 4 ways.
… But need to consider more.
… Friday at 10 there will be another session
Baoping_Chen: Implemented voice communication surveys
… Launched commercial surveys
… About 5 million users expected in 2019
… Sign language recognition
… Community group.
Shigeru: Fair use
… Use cases
… Conclusion is that I will contact experts for more uses cases.
JNovak: DNT:
… WG ended, CR will become a Note.
… Try to find privacy-friendly alternatives
… W3C still cares.
… Web advertising CommGroup continues.
Michael: Domain-specific languages:
… Examples: Math, Chemistry.
… They need semantics and a11y.
… Houdini may help.
… Use cases from publishing showed convergence.
… Horizontal review breakout:
… Challenges, how hard it is, how long it takes.
… At what time to do the review.
… Github workflow issues
… Self-review
Zhiqiang_Yu: Quick apps:
… Combine portability and native efficiency
… API and memory footprint
… We looked at details, uses cases, motivation
… This topic raised a lot of interest.
… Make more capabilities in quick apps.
… Push for XR integration
… AI, low and high level
… High level, such as face recognition.
… New community group recently formed
… Another potential step is work with Web packaging
… Easier distibution of app
… Welcome all collaborations
Kinuko: Web packaging:
… Lots of different groups showed interest, e.g., ebooks.
… How can people get involved?
… Web packaging currently designed for short, caching. Can probably make some other use cases..
David_Singer: Speaking with my AB hat on.
… we had a lively session on evergreen standards
… apparently we need to do registries here (at W3C)
… we wondered whether we could use evergreen standards for maintenance
… also wondered whether we need IPR commitments for those specs
… there's also a question about review...tension between early review v. infrequent review; we want to bridge gaps
… we can use tooling / automation to our advantage (e.g., diffs)
… next steps - there is a wiki page for use cases and aspirations
… we will take discussion into the Process CG that EVERYONE SHOULD JOIN!!
Eric_Eggert: We talked about WAI resources
… videos
… evaluation tools
… WCAG quick references
… we also redesigned the WAI Website
… we also plan to revamp our business case for accessibility documentation
… we plan to ask you for translations of these materials.
… thanks to all (4) who came to my session.
Sandro_Hawke: I ran the credible web session. We hate the term "fake news"!
… we typically say "misinformation"
… we had good turnout but some were not there due to Misinformation conference in London
… focus is how people can trust what they find online; how can we help people being misled
… we aim to create an ecosystem of tools that can assist in evaluation of credibility
… e.g., warnings or annotations about content
… someone even suggested (as a joke) spooky music on risky pages
… question in our session on role of browsers in enabling evaluation tools to plug in
… we welcome feedback on our CG report and also our spec
… next steps: We meet Friday this week; please review the stuff at credweb.org !
Tzviya_and_Angel: We ran a session with Natasha Rooney on positive work environment and effectives task forces
… one thing we did was update /Guide with more information useful to chairs; also overhauled it
… there's also a process doc for busy people
… and a chair buddies option available
… there were many suggestions for outreach to members especially for new members
… some suggestions for making specs easier to digest
… and how to write them
… time zone challenges....
… we welcome participation and suggestions for the working effectiveness task force.
… e.g., internal tooling for GitHub would be welcome if you have internal tools
Angel: Regarding PWE, we have an ombudsperson program and plan to improve it
… we identified two cases from past experience (1) harassment (2) distractions
… to deal with these complicated cases we think the current ombudspeople need to be more empowered.
… we also plan to call them "ombuddies"
… one suggestion we heard was to publish a calendar feed to help people with calendaring
… there is an ongoing review of the code of conduct document; we want some review from diverse cultural perspectives to avoid it being misused.
… we also need some additional cultural sensitivities (e.g., around assertion of one's expertise)
… we would also like to increase support (including monetarily) to help people participate who would not otherwise be able to.
… we have a PWE CG; please join, and give feedback on the revised Code of Conduct
Rob_Smith: I chaired a session on video metadata cues supporting geolocation
… I am leading work on video/geolocation sync ... we have identified requirements for emerging markets
… see @@ for demos in action
… question for the breakout session was whether we could piggyback the html text tract channel for geolocation metadata
… or do we need a special kind of cue for maps
… we plan to resolve this in the WICG
… please join us there.
… many thanks to the media and entertainment group and others who contributed to the discussion
… we'll report the results of this session to the Spatial Data IG and the CG. Thanks!
Francois_Daoust: Gaming on the Web! In 2011 W3C organized a workshop where we identified a number of technical gaps:
… - orientation
… - game pad
… - pass params by reference cross workers
… led to the games CG
… that CG continued to work on use cases and requirements, which went to WGs.
… good news: the gaps that had been identified have been filled!
… of course things have changed since then
… we have more work underway: WebGPU, WebVR...other capabilities emerging to support games on the web
… gamers are a tough audience for specs. They want perfection!
… looking at games is a good way to achieve more "professional" results
… with that in mind, we looked at some remaining gaps
… should we
… - organize a second workshop?
… - revive the CG?
… it was exciting to see motivated people. Thanks!
Kaz_Ashimura: Demos! Come to CSS meeting.
Yoav_Weiss: Ran a couple of sessions. One was on privacy / fingerprinting. We distinguished "mechanism" from "features exposed". We concluded that the mechanism was not harmful.
… there's open debate still on the features themselves
… second session on the exposure of resources; we discussed unexposing a couple of features (e.g., subresource timing)
… the conclusion was @@.
… but we plan to reconvene shortly.
koalie: Thanks everyone! Next up - a reception from 6:30-9pm in the lunch space.
… thanks also to the scribes.
ADJOURNED.