RE: Question #2 on 4.1.1 Parsing - what it covers - ARIA? and How? -and SVG

Aria question.

What I see a lot of is aria issues where the developers use label for (html) then attempt to also use aria-label, aria-labelledby and/or aria-describedby but since there is a hierarchy of use, some override or superscede each other we see things that look right but do not announce right as the aria-labelledby for instance will override a label for and thus the announcing is not what the intended coding was attempting to do. Also, even things as simple as label for ids that do not match the intended input field. Or are for ids that do not exist or are misspelled.

I would presume these all come under this discussion as well.

Alan


Sent from Mail for Windows 10


From: Alastair Campbell
Sent: Monday, February 1, 2016 6:30 AM
To: David MacDonald; Katie Haritos-Shea GMAIL
Cc: GLWAI Guidelines WG org
Subject: Re: Question #2 on 4.1.1 Parsing - what it covers - ARIA? and How? -and SVG

David MacDonald wrote:
> Strictly speaking malformed aria would be limited to 1.3.1, or 4.1.2 (or perhaps 1.1.1 on image or
>  input name messups.)
> Not what I would want... would love to include in 4.1.1 but it is not specified there. Wai-aria wasn't around yet. 

I think it’s clear that a mis-spelled attribute (e.g. ARIA) would affect ATs in a similar way to a mis-spelled HTML tag, so ideally we could add to or change the SC so that “elements and attributes are complete and nested according to their specifications”. 

Not sure how given the process, but it would be helpful.

I agree that it would trigger issues under 4.1.2 so it is not a catastrophic issue, but as Gregg said "4.1.1 was intended to protect AT against errors  (or sloppy code practices)”. ARIA is exactly the sort of thing a developer wouldn’t notice as it isn’t visually apparent in regular browsers, so it would fit better there.

Jason mentioned that ARIA was "was around and under active development from 2004-2008”. It was around as a spec, but it was not implemented widely enough for there to be evidence of those type of errors affecting ATs, which was the basis for things included in 4.1.1.  I think there is now.

-Alastair