Re: A "plug/play" URI/IDREF proposal

I think you'll be glad if you go with this.

I'm not wild about "3. Restrict...", but it's acceptable,
especially given how valuable I consider 1 and 2.

> Date: Thu, 10 Feb 2000 10:07:31 +0100
> From: Gregor Karlinger <Gregor.Karlinger@iaik.at>
> 
> "Joseph M. Reagle Jr." wrote:
> 
> > [...] I understand Eastlake (or others) might try to post a
> > plug/play proposal on URI/IDREF [...]
> 
> I have followed up the latest discussion about the URI/IDREF problem and
> would like to present a proposal which is slightly different from what I
> have posted in [1] and was referred to by Donald in [2]:
> 
>   1. Remove the 'IDREF' attribute from the Reference element.
> 
>   2. Allow the 'URI' attribute of the Reference element to be of type 'uri'
>      as defined in XML Schema Part 2 [3]. This means the 'URI' attribute is
>      a 'URI-Reference' and can have a 'fragment' part.
> 
>   3. Restrict the semantics of the 'fragment' part of the 'URI-Reference'
>      to a single case: Only allow the XPointer 'bare name' shortcut as
>      defined in [4]. In other words: The content of the former 'IDREF' attribute
>      is presented as the fragment part of the 'URI' attribute.
> 
>   4. All other means of selecting parts of an XML document can be expressed by
>      XPath/XPointer transforms.
[...]


-- 
Dan Connolly
http://www.w3.org/People/Connolly/

Received on Friday, 11 February 2000 16:12:51 UTC