Treegrid Example: Remove explicit role="row" declaration from "TR" elements

[adampage]

Resolves #1049.

Preview revised Treegrid Email Inbox Example

---

WAI Preview Link (Last built on Tue, 10 Jun 2025 20:02:53 GMT).

[adampage]

Hi @mcking65 / @howard-e, it seems the validator is still failing on the absence of an explicit tr[role="row"] declaration.

Aside from that, the updated example appears to work just fine. I’ve only benchtested with macOS VoiceOver + Chrome so far, but the experience is identical with and without the explicit role declaration.

[css-meeting-bot]

The ARIA Authoring Practices (APG) Task Force just discussed PR 3291 - Remove role of row from TR elements in treegrid example.

The full IRC log of that discussion

<howard-e> TOPIC: PR 3291 - Remove role of row from TR elements in treegrid example
<howard-e> github: https://github.com//pull/3291
<howard-e> Matt_King: This is ready for review too but I'm a little confused by what's going on with the validator here
<howard-e> Matt_King: Adam_Page, do you understand why it's still failing here after updating the vnurc even though it's updated
<howard-e> Adam_Page: [describes the failure as "Element “tr” is missing one or more of the following attributes: “role”."]
<howard-e> Matt_King: I wonder why the validator thinks it needs to have it
<howard-e> Matt_King: I thought the validator was throwing an error if it was present before but we had a line in vnurc that said don't throw this error
<howard-e> Adam_Page: is there a verbose option for the log?
<howard-e> howard-e: Unsure but I can check
<howard-e> Matt_King: It's so ironic, right?
<howard-e> Matt_King: at least an older version of the validator said the opposite
<howard-e> Matt_King: So which one is needed?
<howard-e> Adam_Page: just found a note in the vnurc which describes this exact scenario so this really is some weird paradox
<howard-e> Matt_King: Wondering if you took out all the rows in vnurc on row, posinset, tr, how it reacts
<howard-e> Matt_King: And also setsize but less sure about that
<howard-e> Matt_King: if there's still validator bug fixing needed, we should just make sure the right issues are raised
<howard-e> Adam_Page: I will comment out those lines in vnurc for now
<howard-e> Matt_King: I'm checking to see if the prior validator bugs we filed are still open
<howard-e> Matt_King: Ultimately, we want to figure out that if we remove the row=role, what validator needs from us
<howard-e> Adam_Page: Yep. I just pushed a commit so I can check what happens there

[adampage]

Heya @mcking65, it looks like we can remove 4 of our vnurc rule filters, but the “tr is missing role” error still remains.

[css-meeting-bot]

The ARIA Authoring Practices (APG) Task Force just discussed PR 3291 - Remove role of row from TR elements in treegrid example.

The full IRC log of that discussion

<jugglinmike> topic: PR 3291 - Remove role of row from TR elements in treegrid example
<jugglinmike> github: https://github.com//pull/3291
<jugglinmike> Matt_King: We have the role of "row" declared explicitly on the TR. It should be unnecessary, but we put it there because we have "pause" and "set" and "setsize", and the validator complained when "role" wasn't declared
<jugglinmike> Matt_King: Now, it says that "role=row" is missing. It seems like there is still a validator bug.
<jugglinmike> Matt_King: One thing we could do is that we could put new changes into the validator configurations to ignore this error
<jugglinmike> Matt_King: That's probably better because then, at least the example wouldn't be telling people, "hey, you shouldn't declare role=row on your TR elements"
<jugglinmike> Matt_King: Do we land this PR with additional changes to the validator? Or do we not land it until the validator is in a better state? Or is there a third option that I'm not considering?
<jugglinmike> jongund: We're also putting "role=gridcell" on the TD elements
<jugglinmike> Matt_King: That's a good call-out. That should be the default implicit row.
<jugglinmike> jongund: If we're taking out "role=row", I think we should also be taking out "role=gridcell"
<jugglinmike> Matt_King: I agree with that
<jugglinmike> bryan: For someone who's looking at the design pattern, how would they know what the purpose of "role=gridcell" is?
<jugglinmike> Matt_King: We do have a note in the pattern, I think. I know we do in the "table" pattern. We tell people that you don't have to put "cell" on the TD elements
<jugglinmike> bryan: I've seen people put "gridcell" on DIV elements and all sorts of other stuff
<jugglinmike> bryan: Where the div is inside a table
<jugglinmike> bryan: It screws everything up
<jugglinmike> Matt_King: I wonder if we put a note about this in the example
<jugglinmike> Matt_King: There's nothing about implicit semantics anywhere
<jugglinmike> Matt_King: If we added something about that, and if the validator isn't supporting the implicit semantics, then I don't know how well-aligned APG should be with the validator
<jugglinmike> Matt_King: If we put code out there, and say "this is the way to code it", and if that throws validation errors... I'm not sure what the best thing to do is in this case
<jugglinmike> jongund: I think we want to tell people how to code using the best practices we know. If the validator isn't supporting that... I don't think we should limit what we tell authors just because the W3C validator has a bug.
<jugglinmike> jongund: How widely used is the W3C validator used, anyway?
<jugglinmike> Matt_King: I have no idea
<jugglinmike> jongund: I'm guessing it's not widely used, just given the problems we've had with it
<jugglinmike> bryan: I do not use it
<jugglinmike> Matt_King: We don't use it at Meta
<jugglinmike> bryan: I've never heard of anyone using it in my practice
<jugglinmike> Matt_King: I hadn't originally thought that implicit semantics would be in the scope of this pull request
<jugglinmike> Matt_King: I'm leaning away from closing the pull request, at least for now, but it doesn't sound like something we're going to close quickly
<jugglinmike> Matt_King: Okay, that's very helpful!
<jugglinmike> Matt_King: I think the next step on this pull request, as jongund suggested, is to remove "gridcell", and then after that, to configure the validator accordingly
<jugglinmike> Matt_King: I'm going to look at what we did in "table". It's probably the case that we have some notes in different places. I'll try to sort that out. I think that would be the next step after the other steps. But it might be a separate issue and another pull request
<jugglinmike> Matt_King: We'll make a decision; we're not quite there, yet