Re: About tests 37-41 (headers)

These tests were discussed last year but a consensus was not reached. A 
search of the archived postings for this list will reveal the discussions.

There were several arguments for allowing the skipping of header levels 
including Eric Myer's article at:
http://www.meyerweb.com/eric/thoughts/2004/07/21/pick-a-heading/

Jim Thatcher's argument below is also convincing for allowing the skipping 
of header levels.

Ben - Could you describe your navigational bar example?

However, I've heard anecdotal information that leads me to believe that most 
documents benefit from following proper header nesting. Headers do imply 
structure and their misuse, such as using headers for visual formatting, 
creates known problems.

The DRC report [1] found that blind users complained about "cluttered and 
complex page structures". I'm not sure if this includes improper header 
nesting but that may be a part of the problem.

Perhaps, as Johannes mentioned, the proper nesting of headers is only 
important within a structural area.

Cheers,
Chris

[1] http://www.drc-gb.org/publicationsandreports/2.pdf


----- Original Message ----- 
From: "Jim Thatcher" <jim@jimthatcher.com>
To: "'Loretta Guarino Reid'" <lguarino@adobe.com>; "'Gregg Vanderheiden'" 
<gv@trace.wisc.edu>; "'Vicente Luque Centeno'" <vlc@it.uc3m.es>; "'Chris 
Ridpath'" <chris.ridpath@utoronto.ca>
Cc: "'WCAG'" <w3c-wai-gl@w3.org>
Sent: Monday, February 20, 2006 9:37 PM
Subject: RE: About tests 37-41 (headers)


>I don't know what Ben's Navigation bar example is, but I suspect it is
> related to what I want to say. I believe that any restriction on allowed
> order of heading tags is wrong and based on an old fashioned (linear) view
> of a web page as a paper document. But web pages have many levels (areas) 
> of
> structure, Navigation bars, left or right navigation or advertising areas 
> or
> link areas, and, say, main content area(s). Different visually styled 
> "area
> headings" and "section headings" will/should appear in any and all of 
> these
> (perhaps in each area well structured). When you put these major sections
> together, there is no requirement and no predicting how the last heading 
> in
> one area relates to the first in another area. With CSS positioning the
> areas can be in any order. So I think it is reasonable to expect headings 
> to
> appear an any order what so ever.
>
> Jim
>
> Accessibility Consulting: http://jimthatcher.com/
> 512-306-0931
>
> -----Original Message-----
> From: w3c-wai-gl-request@w3.org [mailto:w3c-wai-gl-request@w3.org] On 
> Behalf
> Of Loretta Guarino Reid
> Sent: Sunday, February 19, 2006 8:25 PM
> To: Gregg Vanderheiden; 'Vicente Luque Centeno'; 'Chris Ridpath'
> Cc: WCAG
> Subject: Re: About tests 37-41 (headers)
>
>
> I thought we came down against requiring strict application, based on 
> Ben's
> navigation bar example.
>
>
> On 2/19/06 6:17 PM, "Gregg Vanderheiden" <gv@trace.wisc.edu> wrote:
>
>>
>> Good work on codifying this.
>>
>> There was a question though about whether this strict an application of
> the
>> headers was required by the success criterion or even by the HTML spec.
>>
>> Does someone remember where we came down on this at the meeting?
>>
>> Gregg
>>
>>  -- ------------------------------
>> Gregg C Vanderheiden Ph.D.
>> Professor - Ind. Engr. & BioMed Engr.
>> Director - Trace R & D Center
>> University of Wisconsin-Madison
>> The Player for my DSS sound file is at http://tinyurl.com/dho6b
>>
>> -----Original Message-----
>> From: w3c-wai-gl-request@w3.org [mailto:w3c-wai-gl-request@w3.org] On
> Behalf
>> Of Vicente Luque Centeno
>> Sent: Sunday, February 19, 2006 6:23 PM
>> To: Chris Ridpath
>> Cc: w3c-wai-gl@w3.org
>> Subject: About tests 37-41 (headers)
>>
>>
>> Hi,
>>
>> I have improved the formalization rules for headers. The following XPath
>> expressions address all headers that skip a downward sequence, according
> to
>> tests 37-41.
>>
>> For all of them (except for h1):
>>
>> * The preceding header is calculated. If none, current header is badly
>> placed.
>>
>> * We take the closest preceding header to the current one (we get that
> with
>> the [1]).
>>
>> * We check if that closest header is OK:
>> H6's closest preceding header must be a h5 or h6.
>> H5's closest preceding header must be a h4 or h5 or h6.
>> H4's closest preceding header must be a h3 or h4 or h5 or h6.
>> H3's closest preceding header must be a h2 or h3 or h4 or h5 or h6.
>> H2's closest preceding header must be a any header.
>> For all: otherwise, current header is badly placed.
>>
>> Comments?
>>
>> The rules are the following:
>>
>> //h6[not(preceding::*[self::h1 or self::h2 or self::h3 or self::h4 or
>> self::h5 or self::h6][1][self::h5 or self::h6])]
>>
>> //h5[not(preceding::*[self::h1 or self::h2 or self::h3 or self::h4 or
>> self::h5 or self::h6][1][self::h4 or self::h5 or self::h6])]
>>
>> //h4[not(preceding::*[self::h1 or self::h2 or self::h3 or self::h4 or
>> self::h5 or self::h6][1][self::h3 or self::h4 or self::h5 or self::h6])]
>>
>> //h3[not(preceding::*[self::h1 or self::h2 or self::h3 or self::h4 or
>> self::h5 or self::h6][1][self::h2 or self::h3 or self::h4 or self::h5 or
>> self::h6])]
>>
>> //h2[not(preceding::*[self::h1 or self::h2 or self::h3 or self::h4 or
>> self::h5 or self::h6][1])]
>>
>> //h1[not(true())] (which is //h1[false()], which is () )
>>
>> Vicente Luque Centeno
>> Dep. Ingeniería Telemática
>> Universidad Carlos III de Madrid
>> http://www.it.uc3m.es/vlc
>>
>> On Thu, 16 Jun 2005, Vicente Luque Centeno wrote:
>>
>>>
>>> All these examples also work with my rules :-)
>>>
>>> On Wed, 15 Jun 2005, Chris Ridpath wrote:
>>>
>>>> These tests are designed to detect when header levels are skipped in
>>>> a downward sequence. For example:
>>>> h1 followed by an h2 is OK
>>>> h1 followed by an h3 is bad
>>>>
>>>> Another example:
>>>> h3 followed by an h4 is OK
>>>> h3 followed by an h5 is bad
>>>>
>>>> Header levels can be skipped in an upward sequence. For example:
>>>> h4 followed by an h2 is OK
>>>>
>>>
>>>> On Wed, 15 Jun 2005, Vicente Luque Centeno wrote:
>>>>> Those h2 having no preceding h1 are an error:
>>>>>
>>>> The group does not have a test for whether the first header in the
>>>> document must be an h1.
>>>
>>> Neither of my expressions say that explicitly. However, by proper
>>> deduction and combination of my rules, you will implicitly obtain that
>>> there is no other possibility. So, yes, the first header in the
>>> document must be an h1 (or there should be no header at all) if tests
>>> 37-41 should be followed. There is no explicit rule for that, but it
>>> is derived from tests 37-41.
>>>
>>>> There are no tests that check for multiple h1s in the same document.
>>>
>>> That's why I did not include a rule saying:
>>>
>>> count(//h1) <= 1
>>>
>>> Cheers.
>>>
>>> Vicente Luque Centeno
>>> Dep. Ingeniería Telemática
>>> Universidad Carlos III de Madrid
>>> http://www.it.uc3m.es/vlc
>>>
>>>
>>
>>
>
>
>
>
>
>
>

Received on Tuesday, 21 February 2006 15:05:25 UTC