See also: IRC log
Wilco: from Deque, co-facilitator of this TF
Kathy: from Interactive Accessibility, co-facilitator of Mobile A11Y TF
Alice: from Google, working on UI
Shadi: W3C staff contact for this TF
Song: from NIA, Korea
John: from Microsoft
Romain: from Daisy
Michael: W3C staff contact for parent WCAG
WG
... focus on interfacing with WCAG WG
<rdeltour> ACT Overview: https://www.w3.org/WAI/GL/task-forces/conformance-testing/wiki/ACT_Overview_-_What_is_ACT
Wilco: many tools out there, each slightly
different, conflicting results
... that's a problem
... want to figure out, how to get to a core set of rules that we can
all agree on
Michael: often get questions, can't provide
support requests
... sometimes authoritative review needed
Kathy: sometimes also questions on coverage of WCAG
Wilco: impression that previously such discussions were avoided
Michael: when we get a request for
interpretation, often it is one that the group agrees on
... when not, we look into why
... sometimes unclear with different history of how requirements were
created
... often documentation not available anymore
Wilco: would try to interpret to take the load off the WG
Shadi: so we bring results of our discussion to the WG?
Michael: only in areas of dispute, if existing supporting resources are not sufficient
Wilco: difference between success and failure?
Michael: failure is direct result - if
there is a fail, requirement is not met
... to prove success, not meeting a technique is not exhaustive
... could meet it in a different way
Wilco: we assume accessible unless a failure is found
Michael: might not be a common approach
Wilco: accessibility support is also an
open issue
... in aXe we check if implementation is viable
... and if there are alternatives
Kathy: how do you keep that up-to-date?
Wilco: this is exactly the problem description
[Wilco walks through "Goals" from ACT Overview page]
Wilco: rules will not be exhaustive
Shadi: but developed tests should be authoritative
Kathy: warning about calling it conformance
testing
... can test requirements partially but often not fully
... difficult to prove conformance
... already confusion out there, need to be careful about wording
Romain: need to be careful with the word
authoritative as well
... not exclusive, tool vendors can develop any rules too
Shadi: agree, was looking for the issue of
validation
... somehow recognizing the rules that have been validated for
interpretation of WCAG
Kathy: conformance has very specific meaning in WCAG
Shadi: just wanted to make sure there is
the tie-in to WCAG as a specification
... rather than accessibility testing in the wild
John: layer of conformance - browser, platform, content?
Wilco: it depends
Kathy: even absolut failures are fairly rare
Wilco: not sure I agree with that
... there are always exceptions
... like conforming alternative aspect
... but that is caught in semi-automated
Kathy: ok, as long as semi-automated
Shadi: there is this perception that it is
all about automated only
... but in fact, even most auto-WCAG tests are actually semi-automated
Kathy: confusion out there about what is
conformance really
... most focus is on the success criteria, rather than the actual
conformance requirements
... we propagate that by calling it conformance testing or automated
testing
... maybe one goal of this group is to help define what is needed for
conformance testing
<Zakim> MichaelC, you wanted to note conformance testing only needed for conformance claims, which aren´t required
Wilco: one of the discussions with WCAG WG, also in light of Silver
Michael: one of the requirements for WCAG was to have conformance claims
Kathy: conformance claims becoming very
important in the US, for example
... seeing more and more the need
Wilco: conformance testing important in the Netherlands
Shadi: take away is that we have an issue
with the naming
... suggest to take that back for later discussion in the group
Wilco: had previous discussions on that
Kathy: agree, difficult problem but needs to be discussed
[Goal #4]
Wilco: not limited to HTML
... also CSS, etc
... even UAAG
Shadi: is it a scope creep to include browser testing beyond content testing?
Kathy: think it would be important to keep in mind
Wilco: primary focus HTML, CSS, etc
Boaz: is the framework exclusive to
accessibility testing at all?
... possibly shareable beyond accessibility alone
Kathy: agree with that comment
Wilco: would not want to put a hard
boundary to UAAG
... everything converging
Kathy: agreed, also in light of Silver
Shadi: convinced, just worried about scope creep
Romain: digital publishing technologies not
separate from web technologies
... ePub uses web technologies
... our selectors are pretty much focused on web technologies
[Target Audience]
Kathy: how to treat false positives?
... only rules that don't create false positives?
Romain: think so
Wilco: would say TBD
Romain: could be goal to reduce false positives
Shadi: do we want to differentiate false positives and false negatives, or just talk about accuracy?
Romain: false positives can often block from publication for no reason
Kathy: some tools will suggest additional code to fix a false positive
Wilco: agree need to look at accuracy
Kathy: can't eliminate false negatives in semi-automated
Shadi: depends on atomicity of tests
Romain: initial rules exclusive to HTML,
CSS, and ARIA?
... what does that specifically mean?
Wilco: we're on the Initial Rules section
of the Overview
https://www.w3.org/WAI/GL/task-forces/conformance-testing/wiki/ACT_Overview_-_What_is_ACT
... Romain said it sounds like the Rules development is in scope of ACT
... let's change that
Romain: what's the ACT role? Overseeing or coordination?
Wilco: we should coordinate with auto-wcag
... **changes the section accordingly**
Kathy: we wanted to talk about whether we need to further define "HTML, CSS and ARIA"
Wilco: I think we don't need to be more
specific
... we don't want to promise more that what we're doing to do
Romain: do we want to mention the EPUB
rules
... we will develop a rule set specific to EPUB, ultimately based on the
ACT fwk
... I'm not pushing for mentioning it here, but it might help with the
2-implementation req for the CR
shadi: we don't want to boil the ocen right
away
... but I'm wondering if we should rephrase the second part "These rules
will focus on..." to be more inclusive
Romain: what about saying sth along the lines of "and coordinate with any other community interested in developing rules for the ACT fwk"?
Wilco: I want to be explicit about Auto-WCAG
Romain: +1
Shadi: ok for Romain's proposal, with s/any other communry/other groups/
<shadi> +1
Wilco: "Use Existing Rulesets"
... we want to coordinate with orgs that have rulesets, and help them
transition to ACT
kathy: we should also look at orgs checklists, not only automated tools
Wilco: **opens the editor to improve the
language**
... add "or test procedures"
<Zakim> shadi, you wanted to ask about "will work with"
shadi: "will work with" means we'll support organizations?
kathy: what about we change that to say we're going to use their rulesets
shadi: there's the aspect that we're going
to look at their stuff
... then there's the question of how to assist people who want to
migrate
... provide some kind of support or guidance in getting their rules in
the right format
kathy: then let's say that explicitly
... "The ACT TF will collect the rulesets from organizations interested
in contributing to the report, for the purpose of forming and evaluating
the framework"
Romain: do we need the agreement from these organizations? we can just pick the rules
kathy: right, but some rules are not open, so those organizations need to explicitly contribute them
Wilco: we also want to say that the orgs can contribute their rules to Auto-WCAG
kathy: the second thing is to talk about
the "specific acceptance requirements"
... do we need to mention the orgs that do not want to contribute?
Wilco: it was there to address people asking "why should I contribute?"
kathy: ok, we might need to rephrase that
shadi: "ACT TF will provided guidance to help organization migrate their test rules and procedures"
Wilco: we're supporting auto-wcag
shadi: auto-wcag is not a body that can help organizations, it's the other way around: organizations who want to contribute can join auto-wcag
wilco: **rewords the paragraph accordingly**
eric: we should use the present tense
sharron: you need to be persuasive
eric: we can list the benefits of contributing
shadi: I'm wondering if the promotion part
should be in another document
... we use the words "ruleset" "test rules" "test procedures"
... are we clear about what these are?
sharron: do you say we need a glossary?
shadi: we might be able to use one term
eric: for me, a ruleset is scoped to the thing to test, like "ruleset for WCAG AA"
wilco: in our terminology, all of these are rules, whether automated or not
shadi: for me, rules almost means automatated, test procedure is more vague
wilco: a test procedure is part of rule
kathy: for our existing ruleset it doesn't
really matter
... we're giving examples, which kind of defines what we mean
romain: it's an overview document, we may not need to be super-specific
<shadi> [[Rules is the overall term, Test Procedures is part of a Rule]]
eric: as someone from the outside, it might be good to define the terms
shadi: we should revisit the wiki pages and make sure that we're consistent
wilco: do we want to list the deliverables in this document?
shadi: yes, it fits well
eric: I like being able to learn this info,
but it can be hard to keep in sync with the formal deliverables
... maybe we can just link to the deliverables doc?
shadi: the deliverables are quite stable
sharron: why are there 2 groups?
shadi: to maximise the likelihood of contributions
romain: you don't need to be a w3c member to join a CG
shadi: it's also a source of confusion,
people do not always know which group they should join
... people are sometimes confused by the whole w3c structure: TF, WG,
CG, etc
romain: the description of "Benchmark
method" sounds confusing
... it's a description on how to test for accuracy
shadi: there's accuracy and validity, two different things
wilco: validation should be part of the ACT
fwk
... when you write a rule you need to provide enough background
... how do we validate the rules that can go in the ACT rule suit is
still an open question
shadi: I always thought that would be part
of the benchmark method
... what else can go in the benchmark method?
... do we provided a litmus test for a specfic rule
wilco: yes
shadi: so it's only the accuracy check, not
the validity check
... the validation needs to happen on a rule basis (or classes of rules)
<shadi> [[validity vs accuracy checking]]
romain: in my understanding the validity
check is done manually when writing the rule.
... the benchmark tool tests only the accuracy
wilco: what if the validity changes later
shadi: yes, the AT can change, etc
wilco: rules will need to be updated as
technology change
... it doesn't mean the rule was invalid
... but new technologies will need different rules
... a rule is valid given its scope
... this doesn't change over time
... whereas the accuracy may change depending on how AT or UA evolve
shadi: we get the question of validation
very often
... how does it move to a contributed test to become approved
... we should describe it somewhere
... if one of the thing we need to think of
romain: we're talking about incubation
... do we need to formalize the process?
shadi: at least some description to give
the rough idea
... we envision there will be a process for approving test rules
... it is kind of missing in the overal description
romain: if the benchmark is to test the accuracy, does it mean we lack a deliverable to describe the validity process?
shadi: not necessarily a deliverable, it can be another section in this list
kathy: WCAG has to review the work for the
TF
... the TF has to define what the rules are, then the WCAG reviews the
TF work
wilco: the ACT TF can get consensus on rules
kathy: there is a document on the WCAG wiki
that's institution knowledge
... do we have WCAG people in the TF?
shadi: we didn't want to pull for the WG, but bring new people
kathy: having somebody in the TF will help with the amount of back and forth
wilco: I think it will be straightforward enough
shadi: we need to have bridges
kathy: I may be willing to help out
... it will make things faster within the TF, at least to identify the
things that need to go back to the WG
wilco: **edits the deliverables overview**
romain: "results sufficiently accurate for (non) conformance testing to WCAG" sounds like a bold statement
shadi: reword into "collection of rules that have passed the validation and benchmarking requirements"
kathy: in the first bullet, I would remove "conformance". we're talking about test rules for accessibility testing.
<jcraig> https://webkit.org/blog/3302/aria-and-accessibility-inspector/
<jcraig> aboxhall: demo similar to jsbin layout
<jcraig> aboxhall: html and js visible, rending and run button with console output on the other side
<jcraig> aboxhall: describes code sample that will changes the label property of an accessibility backing node for a DOM element
<jcraig> aboxhall: runs voiceover with standard labels.... clicks run... console prints done. voiceover reads new label on the button.
<jcraig> aboxhall: describes similar layout of a slider example with an "onaccessiblesetvalue" event handling silder increment/decrement events
<jcraig> aboxhall: js fetches the node, adds event listener, receives the event (when VO increments the slider) and changes the aria-value/min/max, etc
<aboxhall> One more time with feeling: https://discourse.wicg.io/t/contributing-the-accessibility-object-model-specification/1702
<jcraig> https://github.com/a11y-api/a11y-api/issues
<jcraig> aboxhall: inspecting the lang drop down
aboxhall: ...
partial ax tree in chrome ax extension
showing both unknown role and aria-equivalent role in the inspector
<aboxhall> chrome://flags/#enable-experimental-web-platform-features
set the experimental flag in chrome settings for this tool
then you can request computed role and name (label)
<aboxhall> aboxhall@chromium.org
wilco: feedback on the requirements was
pretty good
... Charu provided feedback
<shadi> https://lists.w3.org/Archives/Public/public-wcag-act/2016Sep/0003.html
wilco: if rules are pseudo code then not a
rule - we have gone beyond this discussion and will not be write pseudo
code
... discussion of the comments
... defining standard common standard - we will make that change
... maps to an underlying requirement - when you write a rule you should
be aware of validation
Romain - sounds like requirement for the rules not the framework
Wilco: yes, it is the ACT framework
Romain: the TOC should be updated to reflect this
Shadi: agree with Romain
Romain: i proposed changes in my email
Wilco: we will get to that when we review the feedback
Romain: we may not want to spend too much time on this until we look at the proposed changes
Wilco: skip over some of the comments from Charu
Move to Romain's comments
Romain: many editorial changes
... see the email for details
<shadi> https://lists.w3.org/Archives/Public/public-wcag-act/2016Sep/0004.html
Wilco: want to be careful with scope
James: how about discovering failures
Romain: next comment
rule will not test the policy itself
Shadi: I don't think we should promote other standards and policies.
James: we would want checks for example for ARIA
Wilco: good for organizations to go beyond WCAG
James: we insist that all tables have captions
Romain: instead of standards, how about guidelines
Shadi: test rules for WCAG 2 and other
standards
... says the same thing in the paragraph
Wilco: I will do some wordsmithing on this
Romain: next comment
Shadi: agrees with the direction
Romain: might be a bold statement. How about "The ACT TF aims to increase the adoption of sane accessibility testing principles."
Wilco: how about promote
... we want to make it easier for developers to come to a shared
approach
James: change should to can
"similar to how the HTML standard is a standard on how to write HTML documents"
Romain: I see what you mean, but I would
avoid the comparison given the very different nature of the two specs
... I would remove it
Wilco removes the part of the paragraph
Romain: /ACT Rules/The rules written in conformance to the ACT Framework specification (later referred to as ACT Rules)/
Wilco makes the change
Romain: s/, that can be used in ATTs or for QA testing of accessibility./. They can be implemented by ATT, or used as a reference when performing manual accessibility QA /
Wilco: earlier we said this was not for QA
... removed the word QA
Wilco: comments from Alistair
https://lists.w3.org/Archives/Public/public-wcag-act/2016Sep/0006.html
Alistair: not throw out all techniques
wholesale
...some may be relevant
...techniques are what people use
Wilco: we should come with a proposal for
the WG on relationship between techniques and rules
...think specifically failure techniques may be an issue in the WG right
now
...relates to our discussion
Shadi: agree not throw away everything
...but think minority will overlap with what we need
...techniques have an indirection different from the rules
...but need to develop rules first then figure out mapping
Alistair: think use techniques as starting
point
...techniques are what people try to implement
...rules should check for that implementation
...rather than to try to test everything possible
Wilco: what happens when people write rules for techniques that don't exist?
Alistair: people should only write rules for existing techniques
<Wilco> ACTION: Wilco adds techniques proposal to the agenda for next meeting [recorded in http://www.w3.org/2016/09/22-wai-act-minutes.html#action01]
shadi: I think someone should work on a proposal, then discuss it in ACT before involving WCAG
wilco: maybe setting up a couple paragraph
to summarize our thoughts, I'll do that
... for the last 15 minutes, let's switch back to Romain's comments
https://lists.w3.org/Archives/Public/public-wcag-act/2016Sep/0004.html
Romain: let's look at the new titles
proposal for section 2 "Requirements"
... current titles sound like reqs on the Rules, but this doc is on the
framework
... I propose to reword the headings
... "1. ACT Rules should be clear" could be "Ensure Rules readability"
wilco: not sure about readability. what's readability vs clarity?
shadi: these are more for tool implementors, not for the general public
romain: proposed heading 2. "Ensure soundness"
wilco: what I want for the rules is to be
valid
... I want the rules to get valid results for a requirement
romain: I thout req #2 was more about mapping a rule to a success criteria, not about correctness itself
wilco: what's the difference between validity and soundness?
alistair: I don't think you can "ensure
soundness" without context
... you're trying to say sth along the lines of "non interpretational"
wilco: what I try to mean is "if you ask a11y experts, they would agree this covers the requirement"
shadi: in my mind you mean validity
... I'm wondering if we're not too high level
... the framework needs to have a description for assumptions
... it needs test procedures
... the outcomes needs to be such and such
... this is what the requirements should output
romain: I think you're too specific, we're
not defining rules but the requirements for the fwk
... we need sth along the lines of "the rules should reference some
accesiibility guidelines"
shadi: we're talking about headings, there is content in the sections
alistair: we're not developing sth new,
other industries that do testing have similiar framework
... we should do a little research, find what's interesting in them, and
start from there
... we don't want to reinvent the wheel
wilco: I agree
... with this I want to outline where do we need to put the emphasis
<shadi_> +1 to alistair - look around at existing frameworks to inform our requirements development
wilco: I also agree with shadi that we can
be more specific
... do we want to dive into this even further?
alistair: we should look at other frameworks instead of losing time at word smithing
wilco: can you do that alistair?
alistair: OK
shadi: the req documents should describe
what the framework should be
... here I think we're describing too much what the outcome of the
framework should be
+1
romain: this is very meta, it's almost a req document on a req document
shadi: for instnace, will the framework
support multiple languages?
... do I need to transform the rules into a programming language?
wilco: this is covered in req 2.1
alistair: I read this as rules being expressed as plain english
shadi: so this whole section means the "test procedures must be expressed in plain english"
alistair: yes
shadi: this document is for us, to be a
reference to write the specification
... I think it needs to be a bit more specific than this, maybe less
specific than this
^^ maybe less specific than what I was describing
<agarrison> ISO/IEC Directives, Part 2 - Rules for the structure and drafting of International Standards - has some useful stuff in it.
wilco: we can restart the requirement discussion in next week's meeting after alistair research
shadi: either we can have things useful by tuesday...
alistair: we may have done some research available in WCAG EM
https://www.w3.org/TR/WCAG-EM/
shadi: let's do some research by tuesday, then review wednesday in prep for the meeting
<scribe> ACTION: shadi to lookup relevant part of requirements from WCAG-EM [recorded in http://www.w3.org/2016/09/22-wai-act-minutes.html#action02]
<scribe> ACTION: alistair to lookup ISO documentation [recorded in http://www.w3.org/2016/09/22-wai-act-minutes.html#action03]