Issue 291

[fstrr]

closes #291

[mbgower]

Thanks for this work, @fstrr. It's a significant improvement.

I think it would be an idea to call the HTML equivalent by the name given in the spec ("semantic elements") , or some modification of it like structural semantic elements. I also think it needs to be clearer to distinguish between the two and make it clear that when html exist, the landmarks do not need to be added. They are redundant (and I think the screen readers have matured enough now that we don't need to request they be added?).
Typically, I've seen this discussed as HTML implicit roles versus ARIA explicit roles.

On another topic...

For instance, if a navigation role is used multiple times on a page, each instance may have a unique label specified using aria-labelledby:

This next comment is likely to evolve into a large discussion, but I will note that we have explored whether it is always entirely necessary to label multiple occurrences of the same landmark. For instance, how much weight should we give to hierarchy?
It seems defendable that the two navigation sections in the following situation are distinguishable. Should it really be a failure if the two lack labels?

<header>
<navigation>
<main>
<navigation>

As stated, that is probably best dealt with as a separate issue. i will undertake to create that.

[fstrr]

hey, @mbgower

Thanks for the feedback.

Would "Using semantic HTML elements to identify regions of a page" be a better document title?

The "For instance, if a navigation role is used multiple times on a page, each instance may have a unique label specified using aria-labelledby content is in the current ARIA11 Technique, so has been there for a good while. That said, the new fourth test procedure does move that away from "may". I think it's probably fair that adding a label is best practice rather than a failure, so taking out that new Test and related Expected Result is probably the way to go.

From memory, Internet Explorer needs roles added to semantic HTML for the landmarks to appear (for example: <main role="main">). Every other browser and screen reader combination supports landmarks. I can add some content saying that ARIA doesn't need to be added where HTML can be used.

[mbgower]

Would "Using semantic HTML elements to identify regions of a page" be a better document title?

Yes, I think so @fstrr ! Agree with your other comments to.

[fstrr]

documents updated per your feedback (and some other things)

[mbgower]

@fstrr I have a concern that although you've changed the name of the HMTL5 document, you are using "landmark" primarily in the content. However, I'm not sure anyone would think of the HTML5 equivalents as being "landmarks". I went to change the terms in the article and ran into something of a Pandora's box, in that there does not seem to be consistency of terminology about these elements, even within the HTML5 specification.
Some seem to be called "sectioning" elements, but that term does not cover all of them. The main element for example is categorized as flow content (as are all these things), which is the parent container for sectioning; however it is not listed as a sectioning element. Using "flow" doesn't seem accurate because there are many things that are not equvalents of landmarks which are also in this category.
Another term I've heard is "semantic html", but that is treated much more broadly in the specification (anything with meaning).
I've also heard "structural semantic content" used, but the word "structural" doesn't appear in the specification.
I'm not sure how to proceed or what to recommend!
Here's how I'd written the first paragraph before deciding to just post this comment:

Technologies that support HTML5.

</section>
<section id="description">
  <h2>Description</h2>      <p>The purpose of this technique is to provide programmatic access to sections of a web page. Some HTML semantic elements (or "sections") programmatically identify sections of a page. They serve the same function as ARIA landmarks. Browsers don't expose such tags directly to users, but they they are made available by assistive technology (AT) to users orient themselves to a page and help them navigate easily to various sections of a page. They also can help sighted keyboard-only users navigate to sections of a page using a <a href="https://www.tpgi.com/improving-access-to-landmark-navigation/">browser plugin</a>.</p>

[fstrr]

Moved this PR to a draft. Will continue to work on this to clarify it. I'm trying to avoid the word "section" where possible to hopefully avoid confusion with the section element and also "sectioning content" which, as you point out, doesn't help us here.

Also, as HTML is a living standard and version-less, I'm just referring to it as HTML instead of HTML5.

(As a side note, I'd like to clear up all the outdated references to HTML and XHTML in the files at some point.)

[fstrr]

Does this ARIA Practices 1.1 ARIA Landmarks Example help?

"It is important to understand that many HTML5 sectioning (e.g. main,nav, aside ...) elements by default define ARIA landmarks."

[mbgower]

"It is important to understand that many HTML5 sectioning (e.g. main,nav, aside ...) elements by default define ARIA landmarks."

It depends how technical we want to get. I am not the best person to nail this down, but my undersatnding is as follows... What aria landmarks do is better described as creating an explicit landmark role for the element, whereas the html5 sectioning element creates an implicit role. Maybe by that token we can say that the html sectioning elements create implicit landmark roles? @scottaohara may be able to give us the best suggested wording.

Oh, sorry! I just realized you were quoting from the APG there!! If they're using that terminology, then I guess we can borrow from it.

[fstrr]

There's also H97, which relates to this.

That document also needs a re-write to fix the horizontal scrolling of code examples, multiple typos, and an example that seems redundant (slightly surprised that page got approved). I'm working on that.

[michael-n-cooper]

Looks good overall. A link to the new technique will need to be added to the appropriate understanding files.

[alastc]

Approved: https://www.w3.org/2023/05/16-ag-minutes#item12

@michael-n-cooper - I updated the links in info & relationships to be relative (they were hard-coded to wcag21). I'll check when it is published, but hopefully that works...