<Judy> scribe: Birkir

Judy, I was just going to ask you if we can use the W3C tools in IRC, such as zakim, and all that.

[JB: post-meeting response: https://www.w3.org/community/about/tool/ --note "tools available by default" including IRC meeting management tools]

<Avneesh> High level EPUB accessibility document https://docs.google.com/document/d/1U0ZyVwPCw0IOb0YM0HiPY-j3NH-SgzuGYO-yVnFwTl8/pub

RD: I am Roman Deltour, based out of Rance (insert scribe note, aka tour de France), work for DAISY consortium. We are working on automated or semi automated testing for WCAG conformance, then implement tests specific to epub.

<Wilco> https://www.w3.org/community/auto-wcag/wiki/Action_items

Wilco: Item 1, workshop dates.

AN: We'll take a look at the Doodle poll.

WF: Birkir's is the only test case worked on in the last month (insert scribe note, today actually), we all need to find some spare time and step up a bit.
... Judy suggests bi-weekly meetings on development of W3C proposal. Propose shorter meetings on proposals specifically in between our monthly meeting.

John: Trouble with audio ..

Wilco, suggest April 22nd for a W3C meeting, May 3rd for our next regular meeting, same time.

WF: Note May 3rd is a Tuesday.

JB: Encourage getting the proposal done sooner rather than later. Bi-weekly pace helpful.

BG: Do we all need to be on all meetings (W3C proposal different).

WF: All people do not necessarily need to be present at all meetings.

<Wilco> https://www.w3.org/community/auto-wcag/wiki/Accessibility_Conformance_Testing_for_W3C

AN: Me and Frank both work at FTB, we may split up the meeting participation.

WF: Sent out a first draft of W3C proposal.
... ACT (Accessibility Conformance Testing).
... Start with introduction, why this is important, leads to problem description.

Talk about assumptions e.g. alt="image" is an assumption, there are scenarios where it might be appropriate).

Assumption about a.t./user agent support (differences in ARIA in particular).

WF: Goal 1 (see draft for exact wording).

John: What is a property.

WF: Look at our templates, they have a lot of properties already defined.

Chirp chirp.

JB: People may need more context to discuss.

AN: Concerned about conformance rules wording.
... We can't cover everything, we need human input. Conformance rules might be too strong a term.

WF: Avoid to strong a term, such as compliance.
... Fur work focuses on informing users when they are not conformant. Passing all these rules does not necessarily imply conformance.

JB: Lots of requests for formal conformance procedures.
... We need a more authoritative approach to determining when content conforms.

AN: Discuss better in W3C proposal meeting.

John: What about "WCAG performance" as opposed "conformance".
... Compliance is the low end of where we want to be, inaccessible pop ups are e.g. WCAG 2.0 AA compliant (see webaim discussion).

JB: Earlier we had concerns about tests that go beyond WCAG. Our draft needs to word that clearly.
... May want to include planning evolution e.g. for WCAG 2.1

WF: Need to avoid false impression that if people pass all tests that they are conformant, maybe we should go for inconformance test?
... concerns about confusion associated with the new term.

John, are we talking recommendation document or a work document?

WF: Open discussion.

Birkir: Floating the idea of a concept such as automated tool compliance test.

JB: Important to be clear that automated testing does not give you the whole picture. Helpful to define a framework that would go beyond automated to semi-automated or expert testing.

Birkir: We might want to consider levels of compliance .. "automated test compliant" "user test compliant" "expert test compliant" ..

<scribe> On-going discussion item.

WF: We need to make sure our tests are reliable, accurate, and predictable.
... We need community feedback on the rules, which helps us to assess the quality of those rules.
... 4. We need a process to describe how rules can be implemented, how we register rules, publish rules etc.
... auto-WCAG has some of this, but may need revision based on the standards we define.

AN: Difference between accessibility rules in item 3 and conformance rules in item 1?

WF: The are the same, need change of wording.

After discussion we may consider changing the wording to 'ACT" rules, needs further investigation.

WF: Are we missing anything?

John: Too many rule sets, try to figure where they overlap and where they don't.

WF: We need to try and combine the experiences of the differnt rule sets and design our template so that it might allow for them.

Birkir: We need to start out with the common denominator, WCAG, but set up the process that may allow for flexibility. We need something out there that is useful soon. We need to start by understanding what is a good rule.

JB: Shadi can probably provide more feedback on the format, with hopes that the group can come up with an accepted common denominator for rule sets.
... Need more feedback on existing rule sets (also Shadi may have input on the format). Need a careful inspection of what is available with regards to content and format -- with consideration to what would be most useful as a common reference set.

John: Flashback to discussions we had 10 years ago. We used Eclipse expert system that worked with 5 different rule sets. I can go check on the format we used and see how digestible it is to humans.
... Every html code became a rule statement.

JB: Pull in other folks e.g. from W3C Interaction, experience from browser testing.

WF: Now I see why Judy recommended bi-weekly meetings. (scribe note: Note the time stamp, we are over an hour in).
... Getting to the rest of the agenda.
... auto WCAG members need to focus on rule development. need a standard way of defining rules with regards to properties, quality, interaction with other rules. End up with implementation mechanism and predictable output.

AN: WE can't be more stable than the browser that processes our code. We can think of our work as similar to WCAG techniques.

WF: Need to discuss reliability e.g. aXe's mantra of "no false positives".
... WE need 95% accuracy.

JB: WE need to sort out what rules can be normative. We need to look at the scale of things that can be tested, can get overwhelming quickly. We need something that is authoritative on some level, but also realistic.

WF: We could look at a small scale test based on what we have implemented.

AN: This may involve "some" manual work. (scribe note, put the word "some" in sarcastic quotes")..
... Think of 2 steps, firstly that the theoretical rule is effective, secondly that te tool implementation of that rule is accurate.

WF: Integration tests may be an additional property of our rules.
... I am going to take the feedback from meeting and look at it. Also encourage al of y'all to share the page and comment on it (note from scribe, we need a bit of southern slang in here).
... We hope to turn this document into something normative.

JB: We need to find out which W3C member organizations are willing to be involved in the work. We need concrete commitment for rule set contribution.
... It may require some opensourcification of rule sets to be donated.

RD: Daisy is a W3C member, will very likely contribute to a working group.

<Wilco> https://github.com/auto-wcag/auto-wcag/

WF: Proposes an off-line catch up with RD.
... Review Github stuff, decide what we want to do with outdated stuff.

<Wilco> http://auto-wcag.github.io/auto-wcag/dist/rules.html

WF: Should include all test cases we have designed so far.
... This week, try starting working with Github over wiki.
... For upcoming work, stop being wicket, start being a Git. (add your work through Github pull requests).

<annika> http://doodle.com/poll/etgpz5a55xeycp3f

WF: Last agenda item .. Doodle Poll.

AN: 4 participants, available on all dates.

WF: Maybe host this around W3C workshop in Madrid? I don't think it will work, too much time needed on workshop.

JB: W3C TPAC meeting in Lisbon in 2016.

WF: I am going to pick a date this week.

John: Final thought, Github a good idea, from text towards technical implementation.

rrsagent make minutes