W3C

Accessibility Conformance Testing Teleconference

10 November 2022

Attendees

Present
daniel-montalvo, Helen, thbrunet, ToddL
Regrets
-
Chair
Wilco
Scribe
thbrunet

Meeting minutes

ACT Standup

Wilco: Working on Image of text, will work on that tomorrow, will have update next week

daniel-montalvo: Call for review text assumptions accessibility support. Announcement for work that we're doing on implementation pages and work with ARIA chairs and ARIA related rules

Helen: House ill, not much today, will start on manual test rules

ChrisLoiselle: I am working through task for conclusion on survey orientation transform. Emailed Kathy on one ask on transform property
… working through 1678 made some editorial comments at Kathy's request.

ToddL: I am working on PR 1926, update heading is descriptive. Made changes that were suggested by Kathy and Wilco. Need re-review.

<daniel-montalvo> Tom: Minor changes from Wilco's feedback on the two color contrast rules

Wilco: Need more reviewers on the contrast. Have a spot on the AG agenda on the 29th
… Who here can review that today or tomorrow so that we can send out a call for review

Helen: Put me down

Todd: I can help look at it

Several CFCs

Wilco: A few CFCs went out this week for new rules. 4 of them (see spreadsheet). Got some +1's, but not many. No objections.
… Any concerns?
… Hearing none, let's put in a resolution.

RESOLUTION: Accept new 4 new rules CFCed last week for review for AG

Table header cell has assigned cells

Wilco: Survey from a couple weeks back
… All rejected.
… Q3 - Wilco response, too broad. Empty tables exist on the web. Empty cells can be used for styling. Can have table with just headers. A couple scenarios I think this rule isn't quite right for.
… Tom says rule feels backward. Doesn't matter if headers have data, but rather data has headers.

Tom: We have rules that check that data have headings, and make assumptions for simple tables. Haven't had complaints.

Wilco: We have something similar, but some heuristic scenarios for small tables that are only data / headings

Tom: We require that layout tables have role="presentation"

Wilco: But that's not required by WCAG 2

Tom: We require it because it's impossible to make some of these things work without enforcing that because you don't know

Wilco: I think Tom's proposal is a new rule

Tom: Then I don't think this rule makes sense - not sure why a user cares that a heading has no data

Wilco: One scenario I saw was that the scope was done wrong. All of the headers had scope col instead of scope row, so none of the headers had any association.
… and the data cells can still have headings, just not all of them

<daniel-montalvo> https://www.w3.org/WAI/WCAG21/Techniques/failures/F46

<daniel-montalvo> https://www.w3.org/WAI/WCAG21/Techniques/failures/F49

Chris: There is a failure for using presentation on a data table. There is recommendation that layout has presentation. Some issue with blank cells.

Wilco: Next comment from Kathy. Pass 5 not sure this passes 1.3.1. Fail 1 not sure this is a failure of 1.3.1
… Kathy doesn't think a column header with no data is a failure of 1.3.1
… Remaining issues - 1517

Wilco: Q7 - typo in expectation.
… Applicability needs to establish direct relationship between table and header cell.
… Tom notes roles shouldn't be used on table elements
… Wilco notes that accessibility support is poor. headers attribute, only Firefox.
… The accessibility API does not expose table structures, so I don't think this is a browser thing, but it does seem that Firefox seems to work better, so not sure if this is a browser problem or AT problem.

Helen: I have to admit with JAWS and table support, sometimes people think it's not announcing, but sometimes wrong keys are using.
… When asked to check if supported by JAWS, when I've tested it with the right commands, it is announced.

Tom: I hadn't run into issues with it. Maybe I didn't have the right testcase?

Wilco: I know at least Safari VoiceOver does not support

Wilco: Maybe I'm misinformed - I'll check.

Wilco: Kathy and Tom not sure this is 1.3.1. Wilco thinks too broad.

Helen: I think this needs to be refined before we can determine whether or not it maps to 1.3.1

Wilco: Data goes two ways. Screen readers don't announce data for headers, but rather headers for data. As far as 1.3.1 is concerned, feels like two sides of the same coin.

Helen: Normally goes data to header.

Wilco: I'm surprised by the conversation. Feel like looking at headers have data is a safer assumption.

Chris: Would say if one heading is missing out of for, what's the context?

Wilco: This rule definitely needs work. Should there be a rule?

Tom: Do we need the rule for opposite? Anyone disagree?

Helen: I disagree. I don't believe that data needs to be present. Sometimes data are there to put data in or to show progress. Null is valid data in a table.

Wilco: Right, but should the rule be that if there is data, should there be headers?
… A rule like that is making some assumptions. Assuming that all data needs a header.

Helen: It depends. If it's in a table. If it's a layout table, it needs a header.

Wilco: I think we can rule out layout by saying there are headers (th)

Daniel: May be column empty scenario to pursue.

Helen: May have one blank because it's below row headers.

Wilco: Anyone think that's a bad idea? Hearing nothing, we should have a rule like that. Will put that in as a resolution.

RESOLUTION: There should be a rule that checks that data cells has a header (in a table with headers)

Chris: From trusted tester standpoint, you look for programmatically associated headers, which agrees with what we've been talking about.

Wilco: Next question - what do we do with this one?
… I think rule needs to be refined that it's a table with data, do the headers need corresponding data.

Helen: I don't think it needs to be deprecated, just refined.

Daniel: So we'd refine and then create the other in reverse?

Wilco: Looking to Tom to see if there's any issues with going forward with that.

Tom: I can't think of any complex tables offhand that would have issue, but I'd have to think through it some more

Chris: Looking through tutorial tables. Some scenarios with 0 / null data.

<Wilco> https://www.w3.org/WAI/tutorials/tables/caption-summary/

Wilco: I think some of these help. Like the Rome column. This is a complex table with Studio, Apt, Chalet, Villa column headers, 1, 2, 3 bedroom row headers.. then row sections with Paris / Rome.
… If Rome didn't span, but rather had blank cells... but if the blank cells weren't associated you might catch that

Wilco: I think empty table cells should be ignored

Wilco: I don't think you have have headers of headers.

Daniel: I think it depends on how you move
… If you move by row, you'll get Paris, but not by column.

Wilco: I feel like we'll be reviewing this again.
… I think most of us are in favor of moving this rule forward with refinement. Have a resolution to have a new rule for headers from data cells.
… Not hearing objections, this rule needs a liaison
… Helen would you be open to working on this?

Helen: I'll give it a go

Wilco: If you get stuck, pass it my way

Summary of resolutions

  1. Accept new 4 new rules CFCed last week for review for AG
  2. There should be a rule that checks that data cells has a header (in a table with headers)
Minutes manually created (not a transcript), formatted by scribe.perl version 192 (Tue Jun 28 16:55:30 2022 UTC).