Re: change proposal for <main>: possible validation warning heuristic for misuse

On Sun, Dec 2, 2012 at 12:19 PM, Maciej Stachowiak <mjs@apple.com> wrote:

>
> On Dec 1, 2012, at 3:27 PM, Silvia Pfeiffer <silviapfeiffer1@gmail.com>
> wrote:
>
>
>
> On Sun, Dec 2, 2012 at 9:19 AM, Maciej Stachowiak <mjs@apple.com> wrote:
>
>>
>> <chair hat off>
>>
>> Hello Steve & James,
>>
>> One concern raised by WebKit contributors was that <main> could end up
>> getting frequently misused, and as a result, could harm accessibility.
>> James's conformance rules will slightly reduce misuse, but probably not
>> eliminate it. However, if these plus the current conformance rules capture
>> the scenarios of misuse, how about making <main> not have an implied
>> role=main in these cases? In other words, make it affect implementation
>> behavior as well as conformance. This would significantly reduce the
>> potential downsides of <main>, I think. And in the case where you truly
>> need to violate these rules to mark the main content, you can always use an
>> explicit role=main.
>>
>
>
> Not sure about this one.
>
>
> I think this will be a lot more effective at limiting the harm from
> potential improper use of <main> than a conformance error. A conformance
> error is a discouragement for some authors, but most content is
> non-conforming. Meanwhile, implementation behavior can avoid incorrectly
> identifying the main content even in the face of authors who do not
> prioritize document conformance.
>


Could that also include a rule as to what to do in case there is both a
<main> and a role="main" on the page? While it's a conformance error,
browsers still need to decide which one to expose to AT. So, maybe in this
case it would be best to expose the element with role="main" only?



> Another related thought: if the rule for finding the main content becomes
>> a little more involved based on the above rules, then perhaps the spec
>> could document how to find the main content when no <main> element or
>> role=main is present. This would end up documenting the semi-mythical
>> "Scooby Doo algorithm" for cases when the <main> element is absent. I think
>> a unified "find the main content algorithm" applying these sets of rules
>> would be really helpful to accessibility, it would help make clear why an
>> explicit element is helpful even with a fallback approach, and it would
>> clarify that the two need not conflict.
>>
>
> I agree with this suggestion. I would also like to see "Scooby Doo"
> documented properly. I do wonder, however, since (if?) it is only
> accessibility related, whether it should be in the HTML spec, or in the
> mapping spec of Stever, or in a WCAG spec.
>
>
> I think a "find the main content" algorithm has non-accessibility uses as
> well, for example for data mining tools, or for "readability" style tools
> or browser features.
>

Right. So it should indeed be part of the extension spec, and thus
ultimately of HTML, right?

Silvia.