RE: Processing instructions for validating that a document mayaccess data

> From: Dan Connolly [mailto:connolly@w3.org] 
> Sent: Thursday, 17 March, 2005 11:28
> 
> On Thu, 2005-03-17 at 11:39 -0500, Paul Grosso wrote:
> [...]
> > 
> > At my suggestion, the XML CG discussed this at our 
> > meeting this week.  I've also done some more research
> > on my own.
> [...]
> > Regardless, neither I nor anyone on the XML CG expressing 
> > an opinion in this matter can see any strong technical
> > argument against using a PI in this case.
> 
> Hmm...  it doesn't bother you that somebody might already
> be using <?access-control ... ?> for some other purpose?
> 

Please note that [net certain "personal opinion" comments],
I am trying to consider technical arguments from the
perspective of the chair of the XML Core WG.  It is possible
that there are larger architectural issues too--something
like "XML allows you to do that but, for various reasons,
doing so would cause so much trouble for the Web in general 
that higher powers have determined you shouldn't do it"--but
I am not arguing along those lines.  [Personally, I don't
see such a case here.]

It is a little too bad we didn't allow namespace prefixes
on the PI target.

It might be better if VoiceXML used something more specific
such as <?voicexml-access-control ... ?>.

It would also be allowable to use what look like namespace
prefixes in the names of the pseudo-attributes of the PI;
something like 
  <?access-control vxi:allow="..." vxi:deny="..."?>
[not actually sure what I think of that in terms of good
practice, but again there is no technical argument against
that].

> I think of PIs as kinda like comments... they belong
> to authors (or perhaps local communities), not to W3C.
> 

PIs are a valid part of XML.  They can be used to provide
instructions about how to process a document.  I see no 
technical argument against using them as the VoiceXML spec 
suggests.

> How should consumers figure out what specification applies
> to pi <?foo ?>?

It may not be easy.  It may also not be necessary
very often.  Either things work when you have a given
document to a particular processor or they don't.

> Should W3C start a registry of PI names? (heaven forbid!)

No, it shouldn't.

I suppose it's not out of the question that the W3C could 
decide to allow something of the form:

<?vxi:access-control 
    xmlns:vxi="http://www.w3.org/2005/03/voicexml-access" 
    allow="..."?>

(Note, I have not discussed this with anyone else, so this
may be a completely stupid idea.)

paul

Received on Thursday, 17 March 2005 18:11:23 UTC