W3C home > Mailing lists > Public > public-xhtml2@w3.org > June 2008

Re: ARIA syntax proposal/call for consensus

From: Roland Merrick <roland_merrick@uk.ibm.com>
Date: Fri, 20 Jun 2008 12:29:05 +0100
To: Al Gilman <Alfred.S.Gilman@ieee.org>
Cc: Steven Pemberton <steven@w3.org>, "w3c-wai-pf@w3.org PF" <w3c-wai-pf@w3.org>, wai-liaison@w3.org, public-xhtml2@w3.org
Message-ID: <OFF89A9832.E0DFA473-ON8025746E.003E1BEE-8025746E.003F1753@uk.ibm.com>
Greetings Al, during this weeks f2f of the XHTML2 Working Group we agreed 
that we will incorporate support for aria-* into both XHTML 1 and XHTML 2. 

An author should write something akin to  <.... class="checkbox" 
role="checkbox" aria-checked="true" tabindex="0" ...> whether it be HTML 
or XHTML and ideally this would also be the case for SVG and even other 

Regards, Roland

Al Gilman <Alfred.S.Gilman@ieee.org> 
18/06/2008 16:13

"w3c-wai-pf@w3.org PF" <w3c-wai-pf@w3.org>, Roland Merrick/UK/IBM@IBMGB, 
Steven Pemberton <steven@w3.org>
Re: ARIA syntax proposal/call for consensus

On 18 Jun 2008, at 7:30 AM, Gregory J. Rosmaita wrote:

> AlG, i know you are taking the aria dash issue to the HTCG but i 
> wanted
> to ensure that we are all not only in the same library, but the same
> stack, book, page and paragraph -- so, is there WG consensus on the
> following?
> Point 1: ONLY aria- (aria dash) attribute definitions are included in
> the so-called "null namespace"; languages can include those attributes
> on whatever elements make sense for the host language.
> Point 2: There is no "aria" namespace at all
> i would also like to take the opportunity to mention, since i'm 
> currently
> attending a virtual XHTML2 WG face2 face, that the XHTML 2 Working 
> Group
> is prepared to support the use of aria- in the null namespace on 
> elements in the next draft of XHTML 2, should this be PFWG's consensus
> position.
> thanks for the clarification, gregory (does this need to be discussed
> today or thursday?  the sooner, the better)

** summary: no, I am not going to put that on the agenda as a test for
consensus in the terms in which you stated it.

** details:

Let me explain why I have resisted characterizing this as a 100% done 
and what I consider to be the status of the process in PFWG.

The status of the process is simpler.  The proposal at

has been accepted by the group "to be integrated into the Editor's draft
for further review."  It is our working baseline.

The remaining technical risk in this proposal has to do with things 

- What is the conformance paragraph that goes with this host language
embedding paragraph?  The group has yet to see, never mind approve that
paragraph.  Problems with the conformance section could require 
in the prose of the host-language-embedding pararagraph.

- SVG wants to be convinced that the profile with aria-* attributes in
"no namespace" in SVG is adequately checkable in author tools.  They are
still gathering data and making up their mind about that.  This is a
partner/process risk.  We can't remove that risk in a peremptory call
for a vote.

That said:

On the first point, which I would phrase differently:

*  HTML, XHTML, and SVG will incorporate WAI-ARIA states and properties
as aria-* attributes that, pursuant to the Namespaces recommendation,
have no "namespace URI" associated with them.

This has advantages for both angle-bracket-language authors and script
authors in terms of consistency.  (Also the builders of content 
systems -- Matt.)

I have asked XHTML to treat this as our proposal, admitting that it
still has some technical risk until all the ancillary paragraphs such
as conformance are brought into line with it.

This much has received strong and consistent support within PFWG.

* Second point "there is no namespace at all associated with the ARIA
states and properties:"

This is more dicey.  The question has not been discussed by PFWG in 

There is a use case for "associate your markup with a namespace" as 
stated in the
WebArch and quoted by Dan Connolly in his approving review of our 
with host langauge embedding, is for insertion of the markup *by an 
document* where the host language is unaware of or unsupportive of the
ARIA markup in and of itself.


What the TAG hasn't fully appreciated is that the use case of mashing
up markup vocabularies in instance documents and of partnering markup
vocabularies in host languages are different enough so that just forcing
one into the pattern of the other is not appropriate.

So there is in principle an argument to associate the ARIA markup with a
namespace for document-by-document inclusion in John Random Host 
but that's not the question we are asking of XHTML2 WG.  We want them
to embrace the ARIA markup as a partner, reserve the aria- prefix for
all time so they will stay out of that set of attribute names and cede
the maintenance of that part of the no-name attribute space that works
in their host language to us.

There is also an argument that putting this namespace association in
the spec, given the level of confusion about namespaces and language
extension we find in the W3C, would just contribute to confusion and
cause more trouble than it is worth.

So please, as an individual and Participant in PFWG, read and comment on
the active proposal for host language embedding as regards "there is
no namespace (ever) for the ARIA markup" and comment in PFWG so we can
consider that point discussed and conclude consensus.

And in XHTML WG consider what XHTML wants to do for its own purposes
to embed ARIA markup:

a) include the aria-* names, unprefixed with a specification clause that
refers the reader to the WAI-ARIA spec for interpretation of attributes
that prefix-match 'aria-' (or regular-expression-match 'aria-*).  This
is the current working preference of the PFWG, yes.  It has the 
of consistency between HTML and XHTML for authors, which is the main
argument for it in XHTML so far as I know.

b) include the ARIA semantics via prefixed names supported by a 
declaration.  If XHTML2 WG prefers this, I will make PFWG consider 
this by associating the attributes with a namespace name for this manner
of embedding.  I have to be able to tell the Director we didn't break 
XHTML2 Working Group's arm to get them to go along with "aria-" and 
"no namespace."

c) other TBD.


> ----------------------------------------------------------------
> CONSERVATIVE, n.  A statesman who is enamored of existing evils,
> as distinguished from the Liberal, who wishes to replace them
> with others.         -- Ambrose Bierce, _The Devil's Dictionary_
> ----------------------------------------------------------------
>              Gregory J. Rosmaita, oedipus@hicom.net
>   Camera Obscura: http://www.hicom.net/~oedipus/index.html
> ----------------------------------------------------------------

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
Received on Friday, 20 June 2008 11:29:58 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:30:30 UTC