W3C home > Mailing lists > Public > w3c-wai-ig@w3.org > January to March 2013

Re: Rethinking the necessities of ARIA landmark role "main" and HTML5 <main> element

From: Steve Faulkner <faulkner.steve@gmail.com>
Date: Wed, 27 Mar 2013 11:07:51 +0000
Message-ID: <CA+ri+VmamHRZTJaX0az7KsMUdUm+K5GgrPY-XoU3pYYCX-pxeA@mail.gmail.com>
To: Harry Loots <harry.loots@ieee.org>
Cc: W3C WAI ig <w3c-wai-ig@w3.org>, Ian Hickson <ian@hixie.ch>, JF <john@foliot.ca>
Hi Harry,

and Ian's proposal will form a perfect fail-safe when authors do not use
> role=main or <main>.
>

problem is its not a perfect fail safe I have actually looked into a
heuristic approach and like most heuristics it fails at times. From HTML
data I collected and reviewed [1] I found that exclusion was not a reliable
indicator.

There is no reason why the two principles cannot co-exist
>

of course, bit one principle is a thought experiment, the other is
implemented and used already, I urge anyone who thinks Ian's idea is
worthwhile to define how it would work in practice and get implementers
interested in making it real.

[1] http://webdevdata.org/

with regards

--
SteveF
HTML 5.1 <http://www.w3.org/html/wg/drafts/html/master/>
 <http://www.paciellogroup.com/resources/wat-ie-about.html>


On 27 March 2013 10:50, Harry Loots <harry.loots@ieee.org> wrote:

> Steve
> you're right arguing is senseless...
>
> However, it's worth considering the principle Ian promotes:
> That the UA ignores (the way I understand what he proposes)
> <header><nav><footer><aside><etc> and lands on <main>, e.g., (my
> understanding/interpretation) by using a built-in short-cut key exposed to
> all users.
>
> There is no reason why the two principles cannot co-exist, and Ian's
> proposal will form a perfect fail-safe when authors do not use role=main or
> <main>.
>
> Kind regards
> Harry
>
>
>
>
>
>
> On 27 March 2013 11:14, Steve Faulkner <faulkner.steve@gmail.com> wrote:
>
>> Hi all, this discussion appears to be going nowhere
>>
>> We have landmark semantics that are interoperably supported across
>> browsers and AT, we have evidence to suggest that users find them useful.
>> We have mapping of  landmarks built in to HTML structural elements (in
>> various stages of implementation)
>> We have evidence to suggest that authors understand how to implement
>> landmarks.
>>
>>
>> Then we have a thought experiment from hixie that says hey you don't need
>> those landmarks especially role=main. This idea has been brought up over
>> and over by Hixie (note it was rejected on his home turf at the WHATWG) and
>> never gained any traction, browser implementers rejected it in favour of
>> adding the <main> element ( a number of whom have already implemented it).
>>
>> So we now have a method that works (is supported out of the box by AT)
>> and work is also happening to build upon it to provide a simple browser
>> built in skip to content feature that any user can make use of, so in time
>> the necessity of providing a skip link will diminish.
>>
>> It would therefore seem more productive to be debating other topics.
>>
>>
>> with regards
>>
>> --
>> SteveF
>> HTML 5.1 <http://www.w3.org/html/wg/drafts/html/master/>
>>  <http://www.paciellogroup.com/resources/wat-ie-about.html>
>>
>>
>> On 27 March 2013 08:50, Léonie Watson <tink@tink.co.uk> wrote:
>>
>>> Ian Hickson wrote:
>>> "In the interface I am proposing, there is no repeated questioning. The
>>> user indicates to the software that the user wishes to skip uninteresting
>>> content and jump to interesting content, in a single action (exactly the
>>> same kind of action as is used to jump to a header, or to jump to a
>>> specific landmark role). Then, the user agent skips all uninteresting
>>> content and jumps straight to the content the user wants (the same content
>>> as would be marked with <main> or role=main)."
>>>
>>> >From the user's point of view I think this is right. The phrases
>>> "interesting" and "uninteresting" are too subjective to be helpful, but
>>> essentially a single command that moves focus to the start of the main
>>> content area of the page is the goal.
>>>
>>> >From an implementation point of view I think this is inefficient. It's
>>> more reliable and less process intensive to move from A to Z, than it is to
>>> move from A, to B, to C, to D and so on until all that remains by a process
>>> of elimination is Z.
>>>
>>> So if the goal is to have a single mechanism for moving directly to a
>>> given point on the page, what's the hook the UA uses to make that possible?
>>>
>>>
>>> Léonie.
>>> --
>>> Carpe diem.
>>>
>>>
>>> -----Original Message-----
>>> From: Ian Hickson [mailto:ian@hixie.ch]
>>> Sent: 27 March 2013 02:11
>>> To: JF
>>> Cc: w3c-wai-ig@w3.org
>>> Subject: RE: Rethinking the necessities of ARIA landmark role "main" and
>>> HTML5 <main> element
>>>
>>> On Tue, 26 Mar 2013, JF wrote:
>>> >
>>> > A man arrives at the San Jose airport in Silicon Valley.
>>> >
>>> > "I want to go to the campus" he tells the cab driver.
>>> >
>>> > "The Stanford campus?", asks the cabbie.
>>> >
>>> > [...]
>>>
>>> Could you explain to me how this analogy corresponds to the discussion?
>>> In the interface I am proposing, there is no repeated questioning. The user
>>> indicates to the software that the user wishes to skip uninteresting
>>> content and jump to interesting content, in a single action (exactly the
>>> same kind of action as is used to jump to a header, or to jump to a
>>> specific landmark role). Then, the user agent skips all uninteresting
>>> content and jumps straight to the content the user wants (the same content
>>> as would be marked with <main> or role=main).
>>>
>>> The user experience is _exactly_ the same as the experience possible
>>> with explicit landmark roles. The only difference is how it is marked up.
>>>
>>> --
>>> Ian Hickson               U+1047E                )\._.,--....,'``.    fL
>>> http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
>>> Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
>>>
>>>
>>>
>>>
>>
>
Received on Wednesday, 27 March 2013 11:09:00 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 11:09:01 GMT