W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 2011

Re: 2.4.3, 1.3.2

From: Detlev Fischer <fischer@dias.de>
Date: Mon, 15 Aug 2011 14:42:04 +0200
Message-ID: <4E49141C.6020907@dias.de>
To: w3c-wai-gl@w3.org
sorry to enter this discussion at such a late stage.

I think that all examples in all techniques implicitly encourage design 
patterns. They are read as 'best practice'. Clearly, this layout should 
have been created using CSS, not with a layout table 'repaired' with 

That's why I feel uneasy about having such an example in a sufficient, 
even in an advisory, technique. The use of layout tables is actively 
discouraged elsewhere in WCAG (I believe).  While their use does not 
violate any SC when the linear reading order is meaningful, I personally 
think the bride and groom example is possibly in violation of SC 1.3.2. 
In any case, the example is not 'best practice'. It actively introduces 
a problem that the use of tabindex in H4 is then supposed to repair.

On another note, I think it is highly relevant to cover the rather 
frequent use of scripted tabindex for managing the focus order in 
lightboxes used as modal dialogues. While there are many a11y issues 
with lightboxes, they are everywhere and unlikely to go away. In our 
testing experience, they frequently introduce grave problems for 
keyboard users since authors fail to manage focus order properly (and 
fail to move focus back to the trigger when the lightbox is closed).

What's there in the techniques to deal with lightboxes? SCR21 and SCR26 
deal with insertions that are meaningful content in the normal reading 
order. This leaves us with SCR37; but unlike the sorting example in 
SCR37, modal lightboxes often are not a meaningful part of pages' 
reading order: that's why they are generated and removed on the fly, 
meaning that SC 1.3.2 would be OK on the default page.

One could argue that F85 (Failure of Success Criterion 2.4.3 due to 
using dialogs or menus that are not adjacent to their trigger control in 
the sequential navigation order) rules out this pattern anyway. But I 
think it makes a difference whether content clearly makes sense in the 
sequential navigation order or not. F85 should only reply to the former.

A case in point may be glossary entries. Arguably, inserting them into 
an inline text at the point where they are called up disrupts the 
integrity of content and is worse then having such entries inserted 
dynamically at the end of the source code, displayed in a lightbox, and 
put into the the focus order via JS/tabindex.

In practical terms, many common JS frameworks place modal lightboxes at 
the beginning or end of the source code - sometimes, as we have seen, 
for a good reason. Currenly, I believe there are no examples in SCR 
techniques dealing with this frequent pattern. Managing tabindex of 
modal dialogues with scripts (and possibly having alternative renderings 
of lightbox content if JS is inactive) may be an option for a further 
SCR-Technique. It might also need a qualification of F85 to exclude 
dynamically generated content that is better presented outside the 
sequential reading order.

For H4, maybe there are examples where reading order is clearly OK both 
ways? In which case I agree with Adam Solomon that H4 looks more like an 
advisory technique or might be dropped altogether. A difficult one...


  Am 15.08.2011 08:44, schrieb Loretta Guarino Reid:
> On Sun, Aug 14, 2011 at 11:16 PM, adam solomon <adam.solomon2@gmail.com
> <mailto:adam.solomon2@gmail.com>> wrote:
>     Basically the principle is the same, just that html5 applies
>     tabindex to be global (not just focusable form elements), AND lets
>     us declare a value of -1. So we might have some useful examples for
>     these two differences that we would not have in html4.
> This does open up lots of possibilities. This should be an interesting
> discussion.
>     On Mon, Aug 15, 2011 at 9:11 AM, Loretta Guarino Reid
>     <lorettaguarino@google.com <mailto:lorettaguarino@google.com>> wrote:
>         I think you are saying that because the default tab order would
>         conform to WCAG, there is no need for a technique that changes
>         the tab order.
>         But changing the tab order is permissible under WCAG as long as
>         the resulting order "makes sense". And if the author thinks that
>         changing the tab order improves the usability, we want to make
>         it clear that this still conforms. So although the author did
>         not need to change the tab order in any of the H4 examples,
>         changing it in these ways does not violate WCAG.
>         Is there something different about tabindex in HTML5? If not, we
>         may not need an additional technique. We may just need to see
>         whether H4 applies to HTML5, as well, or can be modified so that
>         it is clear that it also applies to HTML5.
>         On Sun, Aug 14, 2011 at 11:03 PM, adam solomon
>         <adam.solomon2@gmail.com <mailto:adam.solomon2@gmail.com>> wrote:
>             I guess I am just missing something. I still feel that
>             sufficient techniques, though not being the only option for
>             satisfying a success criterion, do in fact come to correct
>             some deficiency in the web page. Sure, there are other ways
>             to do it, but the deficiency needs to be addressed one way
>             or the other. In H4, there is no deficiency by WCAG
>             standards. Since the reading order is acceptable (if it
>             weren't, we couldn't use this technique anyway since it
>             violates 1.3.2), the focus order must also be acceptable by
>             definition, and the added benefit of reading one person at a
>             time in the bride/groom example adds no WCAG success to the
>             web page, and addresses no WCAG deficiency.
>             If you all still think that this is a valid technique, then
>             I will draft the html5 technique in accordance with this
>             one, and hopefully have it ready for this week's meeting.
>             On Mon, Aug 15, 2011 at 8:51 AM, Loretta Guarino Reid
>             <lorettaguarino@google.com
>             <mailto:lorettaguarino@google.com>> wrote:
>                 Adam, I am still having trouble understanding why you
>                 think the inclusion of the HTML tabindex technique is a
>                 problem.
>                 I think you are claiming that it is unnecessary? That it
>                 is always possible to use a different technique to
>                 satisfy the success criterion? Am I understanding that
>                 correctly?
>                 Loretta
>                 On Sun, Aug 14, 2011 at 10:46 PM, adam solomon
>                 <adam.solomon2@gmail.com
>                 <mailto:adam.solomon2@gmail.com>> wrote:
>                     Sailesh,
>                     If one would expect to fill out the form one person
>                     at a time, would the default table layout (not
>                     taking into consideration focus order) not violate
>                     1.3.2? After all, the programmatically determined
>                     reading order would read the cells of the table row
>                     by row, not person by person. If so, then this is
>                     not a sufficient technique.
>                     We must then conclude that there is no violation of
>                     1.3.2, and the author's tabindexing is only a
>                     preference, in which a case this technique is
>                     totally irrelevant.
>                     Either way, there is a problem.
>                     On Mon, Aug 15, 2011 at 6:07 AM, Loretta Guarino
>                     Reid <lorettaguarino@google.com
>                     <mailto:lorettaguarino@google.com>> wrote:
>                         On Sun, Aug 14, 2011 at 8:03 PM, Sailesh
>                         Panchang <spanchang02@yahoo.com
>                         <mailto:spanchang02@yahoo.com>> wrote:
>                             Loretta,
>                             In principle, if you content:
>                              >But the use of H4 is not required for SC
>                             2.4.3...
>                             Then why is it listed as a sufficient technique?
>                         Because it is sufficient. You may use it, but
>                         you may use some other sufficient technique.
>                             Adam,
>                             Well in that example of groom and bride,
>                             without tabindex, one may content that
>                             reading order is meaningful. But if one
>                             navigates across fields row-wise, it does
>                             affect meaning or operation. As I said in my
>                             last email, the intent is not to compare
>                             first names but actually enter data into a
>                             form. I imagine most would want to be done
>                             with data for one person then  input data
>                             for the next. While filling out paper forms
>                             too,I'd complete the form for person#1 and
>                             then person#2 and not fill out first name
>                             for person#1 then jump to form for the other
>                             chap and fill out his first name. That is
>                             not logical. On a Web page the fields may be
>                             placed next to each other visually but they
>                             are meant to be navigated "logically" for
>                             person#1 and then #2. It is not the author's
>                             choice or reading  order... the author is
>                             constrained by layout / design and must use
>                             tabindex (h4) to ensure navigation does not
>                             affect operation.
>                             Sailesh

Detlev Fischer PhD
DIAS GmbH - Daten, Informationssysteme und Analysen im Sozialen
Geschäftsführung: Thomas Lilienthal, Michael Zapp

Telefon: +49-40-43 18 75-25
Mobile: +49-157 7-170 73 84
Fax: +49-40-43 18 75-19
E-Mail: fischer@dias.de

Anschrift: Schulterblatt 36, D-20357 Hamburg
Amtsgericht Hamburg HRB 58 167
Geschäftsführer: Thomas Lilienthal, Michael Zapp
Received on Monday, 15 August 2011 12:42:57 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:34:08 UTC