Re: Validity constraints on <section>

On 22 Mar 2013 01:40, "Robin Berjon" <robin@w3.org> wrote:
>
> On 21/03/2013 15:15 , Léonie Watson wrote:
>>
>> The lockerz.com homepage has 108 regions (screen readers use the region
role
>> mapping to report section elements). That's 216 announcements on a single
>> page.
>
>
> Actually I'm counting 149 :)
>
>
>> One example on the lockerz.com homepage causes my screen reader to
announce
>> the following:
>>
>> "Region"
>> "Region"
>> "Region"
>> "Region"
>> "Region"
>> "Region"
>> "Link graphic W310/image0013620927118302ukw51"
>> "Region end"
>> "Region"
>
>
> Yes, that's exactly the problem I was thinking of.
>
> So far the only negative feedback we've received is that in books or
papers it is common to have something that is logically a section not have
a title, for instance a dedication or an abstract.

I've seen slide templates use <section> for every slide. Now, most slides
do have a heading, but some don't because they are an image only or a
continuation of the previous page or something else. I don't think that's
misuse.

How about something more constructive such as an explicit <outline> element
that creates a table of contents through the outline algorithm.

It could also be a feature that browsers provide e.g. in the context menu
and that is overlayed on top of the page to help users navigate. I'd love
this for the HTML spec!

It would help authors to identify if their markup is useful and thus avoid
poor authoring.

Silvia.

>
> While that's a valid concern, I tend to think that from an a11y it's
pretty unhelpful. Such section are visually distinguished (say with page
breaks or italics) but without the visual queues you have to read them to
know what they are.
>
> Chatting with Steve, we wonder if the rule shouldn't be: a section must
have either a heading, or aria-label(ledby).
>
>
> --
> Robin Berjon - http://berjon.com/ - @robinberjon
>