W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > October to December 2003

Re: HTML Techniques: 'tabindex'

From: Christopher P Brainerd <brainerd@hawaii.edu>
Date: Fri, 14 Nov 2003 22:52:55 -1000
To: Sailesh Panchang <sailesh.panchang@deque.com>
Cc: W3C WAI <w3c-wai-gl@w3.org>
Message-id: <53a35353574f.53574f53a353@hawaii.edu>

I fail to the benefit of deviating from the normal tab order by using 
tabindex to skip the navigation section.

Users of voice browsers are accustomed to quickly tabbing past the 
first 20 or so links, that being the repeated navigation bar.

When a page that has an altered tab order is encountered, and the user 
habitually presses tab to skip the expected navigation bar, they will 
instead be skipping over content links.

You must consider user adaptation when you implement user accomodation.

Chris Brainerd
Instructional Designer

----- Original Message -----
From: Sailesh Panchang <sailesh.panchang@deque.com>
Date: Tuesday, November 11, 2003 12:09 pm
Subject: Re: HTML Techniques: 'tabindex'

> Yes that is a good example of contradiction... skip navigation 
> link and  tabIndex cannot co-exist. I am not sure it is a user 
> agent problem but a matter of logical flow  of how   one should 
> navigate the links. The need for tabindex for links arises chiefly 
> when links are placed in a layout table and  logic dictates for 
> instance that one should read down one column and then proceed to 
> the adjacent column. So tabindex might be a good thing to have.
> So  the HTML technique should  suggest: use tabindex for links on 
> pages where  other navigation controls like skip nav are not being 
> employed. 
> And where it is necessary to use a skip nav link, the author 
> should attempt to arrange the links so as to obviate the need for 
> tabindex. 
>  Sailesh Panchang
>  Senior Accessibility Engineer 
>  Deque Systems,11180  Sunrise Valley Drive, 
>  4th Floor, Reston VA 20191
>  Tel: 703-225-0380 Extension 105 
>  E-mail: sailesh.panchang@deque.com
>  Fax: 703-225-0387
>  * Look up <" target="l">http://www.deque.com> *
> 
> 
> 
>  ----- Original Message ----- 
>  From: Michael Cooper 
>  To: W    3C WAI 
>  Sent: Tuesday, November 11, 2003 2:35 PM
>          Subject: RE: HTML Techniques: 'tabindex'
> 
> 
> 
>  Hi Jens - there are a couple reasons the recommendations appear 
> as they do
>  in the HTML techniques: we aren't always sure in all cases what 
> we should
>  say, and in a case like tabindex where it appears in multiple 
> places in the
>  document, we haven't yet gone through the editorial process of 
> making sure
>  the whole document is self-consistent.
> 
>  The reason the current draft recommends against tabindex for 
> links is that
>  we have been presented with evidence that doing so actually 
> causes problems
>  for accessibility. The favourite example is if you use tabindex 
> and skip
>  navigation links on the same page, then the user follows the 
> skip navigation
>  link, the tabindex could undo the benefit of the skip link and 
> take them
>  back to the top of the page, rather than continue in the tab 
> order from the
>  new location. We consider this a user agent problem, not a 
> problem with the
>  use of tabindex in theory, but a sufficiently widespread problem 
> that we
>  felt it was best to say in so many words, "don't do this for 
> now". Thus,
>  what we want to do really is have one technique that says "don't 
> do this for
>  now and here's why" and another that says "here's how in the 
> future one
>  should be able to use tabindex, but user agents don't support 
> this properly
>  yet so we don't expect conforming sites today to follow this 
> technique". We
>  haven't figured out how to express that yet in the structure 
> we've created,
>  and I guess the current expression isn't working since you were 
> confused.
>  Tabindex for forms is different in part because of editorial 
> oversight, but
>  also in part because we have some open questions about it: In 
> current user
>  agents, is tabindex for form controls in a different "tabspace" than
>  tabindex for links? Should it be? Should it explicitly not be? 
> These may
>  seem like arcane questions, perhaps with obvious answers, but in 
> discussion  we didn't resolve them in the timeline of this draft. 
> Some people have
>  homework assignments to look into this.
> 
>  I would say therefore that your confusion reflects in part the 
> confusion we
>  have. Your thoughts and recommendations will be helpful next 
> time we open
>  the books on this issue. So, don't shoot us down, but do fire 
> away with your
>  thoughts!
> 
>  Michael
> 
>  > -----Original Message-----
>  > From: Jens Meiert [mailto:jens.meiert@erde3.com]
>  > Sent: Sunday, November 09, 2003 8:38 AM
>  > To: W3C WAI
>  > Subject: HTML Techniques: 'tabindex'
>  > 
>  > 
>  > 
>  > Before shooting first I want to ask what purpose the 
>  > 'tabindex' handling
>  > recommendation has, referring to the 'HTML Techniques for 
>  > WCAG 2.0' [1]. In 9.5
>  > [2] it is recommended, not to use the 'tabindex' attribute to 
>  > leap e.g. a set
>  > of navigation links, but in another case (see 14.1 [3]) it is 
>  > illustrated to
>  > use 'tabindex' to change the (tab) order of form elements, 
>  > what I perceive
>  > as questionable (and per se contradictory), not only in 
>  > relation to 9.5.
>  > 
>  > Can anyone of you please explain this to me? -- Normally I'd 
>  > think it should
>  > be recommended just the other way around, thus allowing to 
>  > jump over e.g.
>  > navigation links (because there should also be alternative 
>  > <link /> elements to
>  > represent a site's structure, btw), but not recommending to 
>  > violently change
>  > the order of form elements (where I only see the risk to 
>  > irritate users).
>  > 
>  > So I wonder (from Usability perspective, too), what sense the 
>  > 'form case'
>  > (and its example from 14.1) makes:
>  > 
>  >    <form action="#" method="post">
>  >       <p>
>  >          <input tabindex="2" type="text" name="field1" />
>  >          <input tabindex="1" type="text" name="field2" />
>  >          <input tabindex="3" type="submit" name="submit" />
>  >       </p>
>  >    </form>
>  > 
>  > If I imagine 'field1' represents a user's email address 
>  > (which is displayed
>  > first), ain't it 'curious' being at first asked for my name 
>  > (as e.g. expected
>  > in 'field2'), when starting to access the form via pressing 
>  > 'tab'? On the
>  > other hand I often encountered cases where I really wished 
>  > there was a tab
>  > order allowing me to jump over a site's navigation. But, I 
>  > didn't want to shoot
>  > first... ;)
>  > 
>  > 
>  > All the best,
>  >  Jens.
>  > 
>  > 
>  > [1] http://www.w3.org/WAI/GL/WCAG20/WD-WCAG20-HTML-TECHS/
>  > [2]
>  > http://www.w3.org/WAI/GL/WCAG20/WD-WCAG20-HTML-TECHS/#linkgrou
>  ps_tabindex
>  [3] http://www.w3.org/WAI/GL/WCAG20/WD-WCAG20-HTML-
> TECHS/#form_tabindex
> 
>  -- 
>  Jens Meiert
>  Interface Architect
> 
>  http://meiert.com
> 
Received on Saturday, 15 November 2003 03:56:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:26 GMT