Re: Meeting with SVG, XHTML, WAI people to move forward on ARIA as a cross cutting technology

On Oct 18, 2007, at 01:05, Ian Hickson wrote:

> On Wed, 17 Oct 2007, Anne van Kesteren wrote:
>> On Wed, 17 Oct 2007 16:44:31 +0200, Richard Schwerdtfeger
>> <schwer@us.ibm.com> wrote:
>>>
>>> Net: Due to special treatment of the hyphen and colon characters  
>>> in IE
>>> we agreed that it would be best to focus on using underscore. Doug
>>> will be discussing this with the SVG working group.
>>>
>>> So, use aria_checked as opposed to aria-checked or aria:checked
>>
>> Internet Explorer doesn't treat the hyphen specially. I think the
>> argument was that it was better to avoid the hyphen because it is
>> already used in other attribute names (such as color-rendering).

Right. Doug's opposition to the hyphen had nothing to do with IE compat.

> Wouldn't that be an argument in _favour_ of the hyphen?

Depends on whom you ask and whether you are trying to find a naming  
scheme for grouping the ARIA attributes in a way that makes sense for  
a group of attributes in HTML and SVG or whether you are trying to  
generalize a new namespacing mechanism.

I think we should figure out a clean way to do ARIA in particular and  
not to try to establish a new generalized namespacing mechanism.

Like I said on the telecon, the technical compatibility properties of  
the naming scheme and the spec organization are not the same thing.  
HTML and SVG can delegate the definition of ARIA attributes to a  
separate ARIA spec in the case of aria:*, aria-* and aria_*. These  
specs could well advance at their own pace. SVG doesn't need to  
"adopt" "like 70 attributes"[1]. All the SVG WG would need to do is  
to agree to set aside the aria-* names and say that they are  
specified in the ARIA spec. Versioning is a red herring. We don't  
need a single declared version number that covers the combined frozen  
state of both the base language and ARIA. Moreover, versioning  
arguments in general are a distraction. Versioning assumes that spec  
writers are free to make incompatible changes and use a new version  
number as an excuse. The better way to address this problem is to  
constrain spec writers not to break compatibility and doing away with  
versioning. We shouldn't assume that ARIA 2.0 breaks compatibility  
with ARIA 1.0. In fact, I think the Web will be better off if the  
specifiers of ARIA 2.0 feel they have an obligation to stay  
compatible with ARIA 1.0.

Like I also said on the telecon, I prefer the hyphen over the  
underscore. However, I conceded that as far as DOM and CSS  
compatibility goes, the choice between the two doesn't make a  
technical difference (whereas using the colon does). This doesn't  
make them equally good, though. The hyphen is better from from the  
point of view of keyboard ergonomics as well as from the point of  
view of the aesthetic and consistency considerations pertaining to  
language design.

Doug wants[1] to generalize a new namespacing convention that not to  
collide with the existing attribute name grouping conventions of SVG  
but avoids the problems of Namespaces in XML. I think the underscore  
makes sense if that's the problem you are solving, but I disagree  
with the premise. I don't think we should be solving that problem  
here. (I am not convinced we should solve it at all.) Instead of  
creating a new generic namespacing convention, we should be  
introducing ARIA into HTML and SVG (but spec-wise do it by normative  
reference). In that case, it makes perfect sense for the ARIA  
attributes to start with aria-*, stroke attributes to start with  
stroke-* and repetition attributes to start with repeat-*.

> ...adding a fifth would make the language even more confusing,

Indeed.

> though the first form is used for groups of related attributes like  
> the repeat-* attributes in WF2, and would thus make sense for the  
> aria-* group of attributes).

Indeed.

> Also the underscore looks really ugly. :-)

Also, it is harder to write with the usual input methods.

[1] http://www.schepers.cc/?p=46
-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Thursday, 18 October 2007 11:21:29 UTC