Re: Proposal to change documentation on tabindex to strongly discourage values greater 0

2014-10-16 16:16, Patrick H. Lauke wrote:
> On 16/10/2014 14:09, Marco Zehe wrote:
>> This is a proposal to declare tabIndex values > 0 invalid in the spec
>> and the validator, or at least strongly advise against the use of
>> positive integers values for this attribute.
>
> I'd say making it invalid would be a non-starter, as that would 
> potentially make current and legacy websites invalid overnight, and 
> browsers are unlikely to remove support for that scenario for exactly 
> those reasons too.

Declaring something as invalid in the spec is just deprecation, provided 
that the spec also specifies that browsers shall keep supporting the 
“invalid” feature, as HTML5 generally says about old features that 
declares as “invalid”. So it says “don’t use this” but tells browsers to 
handle it in a specific way. (In reality, HTML5 inevitably defines two 
language versions: the “loose” version, which includes all the “invalid” 
features that have well-defined handling in browsers, and the “strict” 
version,  which contains only the features declared “valid”. It just 
does not use these terms and does not define the “loose” version in a 
systematic way.)

> A usage note in the spec, plus a warning in the validator, is probably 
> more realistic as an option.
>
There is a vast number of old, well-supported constructs that are 
declared as “invalid” in HTML5 and that must be supported by browsers, 
to prevent the world, well at least the web, from collapsing.

So it is not unrealistic as such to suggest that tabindex values > 0 be 
declared as invalid.

It is, however, a proposal that should be rejected. There can be good 
reasons to change the tabbing order. Even though this may reduce 
accessibility and usability, HTML as such is not only for the web and 
not only for documents that everyone must be able to use. Moreover, can 
anyone guarantee that there is never a case for changing tabbing order 
in an accessible web page? (Consider a form with tabular data to be 
filled in, with some application-based reason requiring that the most 
natural order is by columns, i.e. 1st column first, 2nd column next etc.)

As authoring advice, we can say that tabindex in general should be used 
with great caution after careful consideration of the problems, if at 
all. But this belongs to accessibility recommendations or general 
authoring advice, not to the rules of the HTML language.

-- 
Yucca, http://www.cs.tut.fi/~jkorpela/

Received on Thursday, 16 October 2014 13:38:29 UTC