-
Notifications
You must be signed in to change notification settings - Fork 70
Minutes 2021 02 11
Present: Emma, Aron, Wilco, Jean-Yves, Will, Anne, Dan Seagul, Daniel Montalvo, Susan, Carlos, Stanimira
Scribe: Susan
agendum 1 -- Final call https://github.com/act-rules/act-rules.github.io/issues/461 -- taken up [from Wilco_]
<Wilco_> https://github.com/act-rules/act-rules.github.io/pull/1522
wilco: asks for comments today on new rule 1.4.5
agendum 2 -- Rules ready for W3C publication https://github.com/act-rules/act-rules.github.io/issues/1120 -- taken up [from Wilco_]
<Wilco_> https://github.com/act-rules/act-rules.github.io/issues/1120#issuecomment-585812132
wilco: look at rules ready for publican
Jean-Yves: issue 492 role attribute has been merged and ready for review
wilco: 493 from Carlos has no progress
Jean-yves: issue 490 not ready
wilco: won't have time this month issue 435 and 437,
jean-yves: 499, 507 not ready
wilco: label is part of accessible name is ready
wilco: header attr was reviewed today
wilco: link and menuitem are in progress, think we're good
wilco: mary jo mueller is stepping back as member and chair of ACT TG
wilco: result is task force is now another person short, was already too small before.
wilco: conversation about some of task force's work being moved over to community group is happening. unsure what that would look like.
wilco, shadi and carlos are working on proposal to bring to TF and community group, possibly by next meeting.
wilco: this does mean things might change about how community group does things
wilco: don't imagine all goals we have can be done as things currently operate
emma: asks how current task force is working
wilco: says shadi is the main person who brings to task force, explains how it is difficult to become invited member.
wilco: work of getting feedback back and forth to task force from community group is inefficient
wilco: current set up assumed other groups would also be doing work and that hasn't happened
Jean-yves: agrees it can be ineffcient but thinks there is value to current feedback loop, otherwise there is a risk of lack of input from community, fresh eyes add quality to rules
emma: agrees with JY but says TF brings another point of view as they're involved with WCAG as well
wilco: agrees and is leaning to having members of TF join the community group. but there is a lot more to work out. there is value to the second step of review and worth keeping
anne: how many people on TF?
wilco: 6 active inc Shadi & Daniel.
anne: only 3 or 4 people in TF who aren't in community group.
wilco: those people are open to joining c. group
wilco: inclined to have separate call inc TF with community group for 'final call' work
agendum 3 -- With autocomplete, should type=tel be allowed on all numbers? https://github.com/act-rules/act-rules.github.io/issues/1541 -- taken up [from Wilco_]
aron: autocomplete rule currently would fail number fields that have type-tel
aron: believe this wouldn't be a failure as 1.3.5 only requires presence of type attr
aron: assumption of 1.3.5 is that type is correct but was though it would be rare to have incorrect type
emma: this isn
emma: this isn't about verifying if the type is correct? that's a user agent bug
aron: spec says it must be correct type to allow succesful data entry but that's not the point of SC 1.3.5
aron: tested if you can still query with wrong type and it was fine
emma: is it possible to come up with example that doesn't have that bug?
aron: regardless whether or not input field is correctly filled out in chromium it works fine. just a firefox issue
aron: bottom line, you can still query property even if type is wrong and that is point of 1.3.5
wilco: colleague corrected me in saying autocomplete from browser is very useful for pwd which suggests it's not just querying value of ac field but how it functions in browser
wilco: reason for second expecation is that it works in the browser
wilco: if you put type tel on field that doesn't allow it, it's not going to work. it's not going to get auto-filled
wilco: that then adds to cognitive load
will: we test credit unions, autocomplete is very important to pwd if it's not done correctly
will: doesn't really understand what's being asked for? exemption for ios issue?
wilco: yes, generally should there be exemptions for user agent issues
will: with general exemption, this remains even if user agents fix it
anne: shouldn't this be in a11y support
aron: last time checked, it was in there. agrees that it is useful but is not WCAG requirement. wCAG only requires something that's programmatically determinable not correct
wilco: programmatically determinable isn't definable. we must decide that
wilco: we can decide if you create something browser can't understand it is not an exception
emma: is it failing the standard or is browser failing user
anne: clarification, is this about have AC and type tel? if you only have type tel on DOB but no AC, that doesn't have anything to do with this issue
wilco: correct
wilco: you are required to have ac on that
anne: doesn't matter if you have ac or type tel
wilco: 1.3.5 requires the autocomplete on such a field
anne: doesn't seem this adds anything useful for users in her country. wouldn't add any value.
anne: do we need to require for AC feature on type tel?
aron: if it's asking for telephone number, yes for AC. that is the requirement, not about the type
aron: thinks defintion of programmatically determinable is quite clear, something that computer can understand,
wilco: 'programmatically determined' needs to be part of understanding.
wilco: if html spec or aria spec defines this, if type=tel being correct is not in spec, it isn't programmatically determinable
wilco: calls for resolution
will: HTML takes priority over OS. otherwise we'd be approving hack
wilco: testing this is on his to do list
Jean-yves: depending on test, we can explain in background that this is common bad practice
aron: one more question, is anne askign to split this rule?
aron: anne please briefly explain suggestion
anne: implementation tests allows for this poor coding sometimes.
aron: rule fails it on assumption that form is taking user input
jean-yves: agree that we can have two rules one for following spec and one manual rule for verifying correct input type
emma: there are similar rules, check that somtthing has a value and then if it is correct
wilco, resolution to change rule to allow auto-complete off
jean-yves: yes as well as new rule that it is correct
wilco: and that second rule is not part of resolving this issue
anne: expectation 2 of that rule is that it's approrpriate
jean-yves: yes but that is not part of this rule
emma: that needs to be more clear
anne: maybe 'appropriate according to HTML spec'?
wilco: yes defintiions should understandable without memorizing
wilco: aron make sure that is captured in the issue
RESOLUTION: allow turn off auto complete
agendum 4 -- Should all table header cells have assigned data cells? https://github.com/act-rules/act-rules.github.io/issues/1545 -- taken up [from Wilco_]
aron: TF has 2 points in feedback that I don't fully agree with
aron: first point, does not seem to be issue that header cells don't have assigned data cells. should be changed for data cells having assigned header cells
aron: assumption is that there are edge cases that some don't have them and rule is valid is.
aron: second Q, TF asked what 'assigned' means. think this is clear in rule and doesn't want to change
wilco: please comment on issue. am inclined to agree with aron
wilco: our own implementation allows header cells with no data cells, generally agrees
aron: could do this both ways. always going to be assumption made for this rule
jean-yves: easy to add applicability of element that is part of table that contains header and data cells, not tables with no data cells. would remove some of those edge cases
aron: agrees, not being empty should be added
wilco: leave comments in issue
daniel S: no final thoughts
Daniel M: looking forward to how we'll organize the group
anne: glad to be back, would like workshop on process of group/TF
aron: glad we clairfied two new rules
carlos: agrees with resolution
emma: good discussion. suggested to group she met that they should pay attention to AG because they are doing a11y work using AI.
wilco: would love to look into how to test for implementors that use AI
jean-yves: looking forward to newcomer workshop
stanimira: bit overwhelmed hope to catch up next time <EmmaJ_PR> group = Evinced