Re: Role attribute values

On 7 Apr 2008, at 1:56 PM, Mark Birbeck wrote:
> Hello all,
> I've been asked to forward these comments to those involved in ARIA.
> Obviously I am happy to do that, but I would just say that I think
> some of my comments relate to exactly what @role is all about, and
> therefore raise some conceptual issues.
> In other words, I think there may need to be some parallel discussion
> over on the XHTML 2 list, about exactly how the values that are
> 'imported' into the XHTML vocab list should be structured.

[disclaimer: haven't reviewed the archives for any such.  I just
have a few clarifications to suggest, not an exhaustive response.]
> Regards,
> Mark
> ---------- Forwarded message ----------
> From: Mark Birbeck <>
> Date: 2008/4/5
> Subject: Re: Role attribute values
> To: Shane McCarron <>
> Cc: XHTML WG <>, Michael Cooper <>
> Hi Shane,
>  I just read this really quickly so these are not all of my comments,
>  but here are a few quick ones.
>  1. Having 'double names' seems to be redundant, and reduce the
>  potential usefulness. An example would be @role="alertdialog", which
>  can be just as easily expressed as @role="alert dialog". This is more
>  useful from a processing point of view, and also reduces the size of
>  the vocabulary.

Maybe "more useful" and "more burdensome" are consistent.  Our  
doing the processing got us to use the agglutenated names rather than
process @role values as a tuple space.

These compound rolenames are there because we accepted a premise that
the processing model would only require the user agent to handle one
widget role (a role with behavioral connotations) at one time.


If the token matches a role as defined in this specification, use  
this token as the ARIA role and end processing of the value.


The API bindings as presently contemplated would not process  
@role="alert dialog"

>  3. Qualifying a name based on its context also seems to reduce the
>  potential usefulness of a name. For example, just because a  
> 'gridcell'
>  is most likely to be used in a 'grid', doesn't mean that is the only
>  place there will be 'cells'. That causes a problem in the future,
>  because it means authors either have to use 'gridcell' outside of the
>  context of a 'grid'--which would seem awkward--or they would have to
>  invent some new role name for 'a cell'. Yet if we call this thing
>  'cell' instead of 'gridcell' from the beginning, we'll not have this
>  problem.

The gridcell has context semantics in terms of how you navigate to it.

The rules for some future context-unaware cell are unknown.

HTML principle: solve real problems.  The real problem we have is
for cells in a rectangular grid with 2D navigation.

>  4. Some things seem to be oddly named. Why is a range control named a
>  'slider'? That points very much towards a specific style of UI
>  rendering, and doesn't really speak to accessibility. Also, since a
>  'spinbutton' is described as a range control, why not make it an
>  actual range control?

These widgets are more concrete than you are used to thinking.

See remark about "GUI bagggage" in reply to Roland.

>  Also, unrelated to the content, why are some of the properties marked
>  up as having a datatype of xsd:string? If it's to inhibit the
>  rdf:XMLLiteral processing, wouldn't it be better to just use
>  @datatype="", and get a plain literal?

This is a technical difference we will need to understand better
before responding.  That is I'm not convinced we fully understand
the difference between "a plain literal" and an xsd:string.


>  Regards,
>  Mark
>  2008/4/4 Shane McCarron <>:
>>  Thanks to the hard work of Michael Cooper, the current list of  
>> ARIA-defined
>> roles is now included in the XHTML vocabulary document at
>>  While doing this, he and I noticed that there is some duplication  
>> between
>> ARIA-defined roles and XHTML module-defined roles.  This was of  
>> course
>> unintentional.  My proposal would be that we harmonize these lists  
>> as soon
>> as possible - preferably by removing any role values that the ARIA  
>> group
>> thinks are unnecessary.   Since the list came from them in the  
>> first place,
>> and was just updated by them a few weeks ago, this makes the most  
>> sense to
>> me.
>>  This is not urgent, and we can update this document as we need  
>> to.  But I
>> didn't want to lose the thread.  Roland, can you please put this  
>> on the
>> agenda for next Wednesday?
>>  --
>>  Shane P. McCarron                          Phone: +1 763 786-8160  
>> x120
>>  Managing Director                            Fax: +1 763 786-8180
>>  ApTest Minnesota                            Inet:
>  --
>   Mark Birbeck
> | +44 (0) 20 7689 9232
> |
> Ltd. is registered in England and Wales, number 03730711
>   The registered office is at:
>   2nd Floor
>   Titchfield House
>   69-85 Tabernacle Street
>   London
>   EC2A 4RR
> -- 
>  Mark Birbeck
> | +44 (0) 20 7689 9232
> |
> Ltd. is registered in England and Wales, number 03730711
>  The registered office is at:
>  2nd Floor
>  Titchfield House
>  69-85 Tabernacle Street
>  London
>  EC2A 4RR

Received on Thursday, 10 April 2008 17:19:22 UTC