W3C home > Mailing lists > Public > public-forms@w3.org > July 2008

Re: XML Base PER

From: Norman Walsh <ndw@nwalsh.com>
Date: Wed, 11 Jun 2008 12:41:26 +0200
To: Steven Pemberton <steven.pemberton@cwi.nl>, www-xml-linking-comments@w3.org
Cc: "Forms WG" <public-forms@w3.org>, "XHTML WG" <public-xhtml2@w3.org>
Message-ID: <m2vdzo5ket.fsf@nwalsh.com>
Hi Steven,

> This is a comment on the XML Base Second Edition PER, from the Forms  
> Working Group.
> 
> The value of the xml:base attribute is interpreted as a Legacy Extended  
> IRI (LEIRI). The characters allowed in a LEIRI include %x0-1F which aren't  
> all allowed in XML 1.0; maybe a note to point out that xml:base doesn't  
> support all Legacy Extended IRIs but only a subset would be welcome.

Yes, I think you're entirely correct. We will add a clarification to
this effect.

> Consuming applications of the new edition might have failures in  
> attributes expecting only URIs unless the application language is rev'd to  
> both include this new version of base AND upgrade all URIs to IRIs at the  
> same time. It seems like there should at least be a note to point that out.

I don't think that's the case. Our changes with respect to LEIRI are,
we believe, entirely editorial. The range of characters allowed in xml:base
attributes has always been what are allowed in LEIRI in XML:

   From the XML Base 1.0 Recommendation:

   The set of characters allowed in xml:base attributes is the same as
   for XML, namely [Unicode].

The spec goes on to say that some characters have to be escaped in order
to transform the value into a valid URI reference and that remains true.

Or have I misunderstood your concern?

> It is not clear from the spec about what is permitted to be considered a  
> URI in a host language. The spec refers to URIs everywhere, without  
> exactly specifying what qualifies. In particular we hope that XML Base may  
> be used at any point some value is interpreted as a relative URI/IRI, even  
> if that value isn't a URI per se.

I'm not exactly sure what more could be said. But note that the
Infoset supports XML Base, so in an XML context, resolving against the
current base URI should do the right thing.

Any spec that has strings that should be intepreted as relative URIs
and supports XML Base can describe the semantics explicitly.

> It would be nice to have an example of an xml:base with an actual IRI in  
> it.

Yes, also a good point.

> In the list of changes it states "The definition of URI reference has been  
> switched from RFC2396 to 3986", but we couldn't locate that definition.  
> Presumably it should be in section 2.

There are a couple of explicit references to 3986 in section 2, but I
think the substance of changing from 2396 to 3986 was updating the
References section.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | Through space the universe grasps me
http://nwalsh.com/            | and swallows me up like a speck;
                              | through thought I grasp it.-- Pascal

Received on Wednesday, 2 July 2008 16:19:39 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 October 2013 22:06:48 UTC