Review ARIA 1.3

[ghurlbot]

Opened by matatk via IRC channel #apa on irc.w3.org

Due: 2024-02-15 (Thursday 15 February)

Spec review request: w3c/a11y-request#72

Significant changes since ARIA 1.2.

[matatk]

tl;dr: All good IMHO. More detailed notes below. A couple of actions:

1. Report apparent minor bug about aria-description not being included in the list of translatable attributes.
2. Consider whether we should suggest slightly stronger wording around the use (or not) of aria-label, aria-description, and aria-roledescription - having seen quite a lot of inappropriate/over-use of aria-label in the wild. Best place to start here would be to search issues to see if it's been raised before; ARIA WG knows what it's doing :-).

Major feature in this release

11-Mar-2020: Add aria-braillelabel

All good IMHO.

There is a note to deter people from using this unless it's really required, and the extra responsibilities they take on by using it.

13-Feb-2020: Role suggestion added

All good IMHO. This will be great! Being able to link the suggestion with further contextual info will be great too.

There are strict requirements on child elements' roles - this seems OK as it matches similar rules for e.g. <details> and <summary>.

11-Feb-2020: Update aria-details to allow multiple IDrefs

All good IMHO.

There is a naming inconsistency vs aria-(labelled|described)by which are similar ones that also accept IDRefs - however, as the details are purposely supposed to be visible elements on the page, this is understandable. Also, aria-details was added in ARIA 1.1.

11-Feb-2020: Role comment added

All good IMHO. This too will be great. It's good that it's a subclass of article, so that it doesn't get added to (i.e. make noisy) the landmark structure.

16-Jan-2020: Add aria-description

All good IMHO.

Slight concern about over-use (or rather under-use of descriptive text in the DOM for everyone), but I don't think that will be as prominent as with aria-label - and definitely accept there are valid use cases for a flat string attribute for descriptions (as well as there are for the equivalent aria-label attribute).

Wondering whether there should be slightly stronger encouragement to use the ID-referencing versions of both name and description attributes, and slightly stronger discouragement of the flat-string ones, but that is a minor issue.

The point about whether flat strings could make translations harder was raised. There is a section on Translatable Attributes - and this actually shows up what appears to be a bug, as aria-description is not included here. I'm not sure what the current state of translation frameworks is, but the spec does draw attention to this issue, which seems most reasonable.

15-Jan-2020: Role mark: Added

All good IMHO.

14-Oct-2019: Add aria-brailleroledescription

All good IMHO.

There is a note to deter people from using this unless it's really required, and the extra responsibilities they take on by using it.

Wondering if a stronger note against the use of aria-roledescription may make sense, but that is a minor point.

Substantive changes since ARIA 1.2

I've reviewed all of the changes, and have no concerns. There are some particularly helpful changes for both AT vendors, AT users, and implementers here.

[matatk]

@mbeganyi-a11y, @fredrika11y: Further to my comment directly above, item 1 of our potential TODOs (that aria-description is missing from the list of translatable attributes) will be fixed by w3c/aria#2043.

That leaves the second question from the tl;dr bit at the top of my comment (i.e. whether slightly stronger discouragement from using aria-label and others may be needed in the spec). I don't want to be over-the-top about it; ARIA WG knows what it's doing, but am interested as to your thoughts on that.

If we do decide to submit something, we should check the issues in the ARIA repo first.

[nschonni]

https://github.com/search?q=repo%3Avalidator%2Fvalidator+aria-label&type=issues issues would probably another source of were the current guidance gets misunderstood