W3C

Accessibility Conformance Testing Teleconference

13 Aug 2020

Attendees

Present
Wilco, Daniel, MaryJo, KathyEng, Charu, Trevor
Regrets

Chair
MaryJo, Wilco
Scribe
Trevor

Contents


Results of CFC on rule: Form control has non-empty accessible name

wilco: Had a few +1s so this is accepted.
... There is already a PR in cg. I am working on a process to automate creation of the markdown we need for the W3C site.

maryjom: We have 5 rules currently out, though one might be aria. Looking to get mroe rules for ag

All table headers have assigned data cells: https://www.w3.org/2002/09/wbs/93339/ACTTableHeaders/results |

wilco: not convinced tables without td should be treated as a failure
... Might not have non-header cells if data fails to load or something.

daniel: Agree

kathy: I also agree

wilco: Should be addressed in assumptions or part of the rule?
... I think it will be pretty easy to come up with real world example of tables without data cells. So I don't think an assumption would be sufficient.
... it is difficult to determine a layout table from a data table. layout table may not need headers.

<Wilco_> https://www.w3.org/2002/09/wbs/93339/ACTTableHeaders/results

charu: was looking at requirements and techniques and couldn't find header requires data cells.
... I think the requirement is that you associate header and data cells

wilco: It isn't a thing for either header or data cells, it is about the relationship

charu: All header cells do not need to have data cells.

<Wilco_> https://act-rules.github.io/rules/d0f69e#failed-example-1

wilco: kathy comment on failed example 1 conveys table header cell with no data cell in column. breaks assumption on relationship through presentation.

kathy: I think the example has value as a th with no data cell, but may be affected by the previous discussion

wilco: I don't htink it communicates well that the row is supposed to span

trevor: Would it require the colspan attribute?

wilco: Yes and at that point it would pass

charu: The requirements say we associate headers and data. Trying to figure out if assigned is meant in the same way as associate.

wilco: Thats the word that is used in HTML.

charu: Should we go with whats in HTML or in WCAG requirement.

wilco: I think we could use associate.

kathy: I like associate over assigned as well.

wilco: as for why some of these don't use scope, technically they do. HTML has auto-scoping so you don't need to assign scope explicitly.
... It can be added to the background section though since it is a common misunderstanding

kathy: I think if its a single level of header scope isn't necessary.

wilco: It detects where in the table it is and then figures out if it is column or row header
... Trevors comment on changing the description away from table.
... we may need to an an example of table with data cells.

kathy: For the second part of my comment, I think it would be better to check that the data cell has a header cell.
... I understand the difficulty of a layout table vs data table.

wilco: We can flip that around to only apply to tables with header cells

charu: The requirement for me is saying that every cell in a data table should have a header cell.

wilco: Is the opposite not true? Are you against this rule or in favor of another rule that checks the reverse relationship.
... We could have two rules, one starts from the headers and another from the cells.

kathy: My challenge with it, is the first check is if its a layout table vs data table. It sounds like the rule is made for beinn done programmatically, but isn't how I would normally manually test.
... Its just not something I check (the th->td relationship), always looking at td->th

wilco: When I did testing, I checked the headings to make sure they had the right scope and then check the table cells to make sure they didn't have attributes that would mess things up.

charu: Do we have a rule if its a layout table or data table? would that be a good idea

wilco: WCAG doesn't prevent layout tables
... If you have a th element, its either a data table or an invalid layout table.

daniel: I tend to first have a look at the headers and be clear of what they are. Then go through the first row or two.

kathy: If you have a layout table, which someone has included a th incorrectly, I don't think this rule would catch that.
... Would there be another rule that would include a layout table that has included that improperly?

wilco: If we had the reverse rule then we could

kathy: I think this is an okay rule if there is another rule to catch those cases

wilco: We might need to add some limitations of this rule. It might not catch layout table or data cells not being associated with header cells, and those should be checked separately

kathy: Could the assumptions or applicability include that?

charu: I think I agree if we can make the distinction that this applies to data tables then its fine to me. Then the techniques and everything aligns with the rule.

kathy: Asking if we could modify assumptions or applicability to say that the rule only applies to data tables.

wilco: Yes, I think there is an assumption here that we can add.

charu: I think that would make it very clear.

wilco: Fix applicability to make table direct ancestor of header. being included in accessibility tree seems unnecessary.
... Detlev's comment on if a mixed native and aria? could this cause problems.
... Think it should be fine.
... Not going to be more specific in accessibility support section.

maryjom: I know aria keeps track of those things somewhere, though it might just be private to the group.

wilco: Would like to be able to link out to external resources with that data, but too hard to maintain on our own.
... 4 nos, so not ready to publish.

Text has minimum contrast: https://www.w3.org/2002/09/wbs/93339/ACTMinContrast/results |

<Wilco_> https://act-rules.github.io/rules/afw4f7#passed-example-7

wilco: Not seeing many comments. Detlev has comment about being unsure of passed example 7.
... This one shows off a font with a sentence that has all characters and contrast 3:1.

<Wilco_> https://www.w3.org/TR/WCAG21/#dfn-text

<Wilco_> https://www.w3.org/TR/WCAG21/#contrast-minimum

wilco: The argument here is that it is purely decorative

kathy: So the intent is to show what the font looks like, not to have someone read what the content

daniel: What if we put something there that doesn't make sense as reading material and is more clear that its only for visual purposes

wilco: Like putting in the alphabet?
... This makes me hesitant, if we are not clear on what decorative means, then the rule becomes ambiguous, which is not allowed.

kathy: I would think it would need an aria-label or something since it is there for the purpose of displaying the font.

wilco: If that isn't decorative, then it would need to meet 4.5 regardless of alphabet or sentence.
... what do you think makes this not purely decorative

charu: I think the words could be rearranged or changed without changing the purpose of the element.

kathy: Would it make it clearer if it was an image of helvetica text?

wilco: The rule wouldn't be applicable in this case.
... We needed something that was text nodes that shows purely decorative text

maryjom: an example I have seen is when they show the font styles and it isn't what they say, but about how they look

trevor: agree that I think its decorative, but am imagining that it could lead to some weird user experiences.

wilco: Unable to add any labelling to something that is decorative
... May need to contact WCAG about the note being wrong.

kathy: whether its decorative or not, how does that affect that it is text or non-text content?

wilco: I think we will need to come back to this next week.

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version (CVS log)
$Date: 2020/08/31 11:15:14 $