Re: Turtle: Prefixes now can contain colons (":")?

Am Montag, den 16.07.2012, 20:43 -0700 schrieb Gavin Carothers:
> On Sat, Jul 14, 2012 at 1:40 PM, Mathias Hasselmann
> <mathias@taschenorakel.de> wrote:
> > Hello,
> >
> > When adopting code for the updated grammar I've noticed that that
> > addition of colons to local names also injected colons to PN_PREFIX:
> >
> > http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/index.html#grammar-production-PN_PREFIX
> >
> > [167s]  PN_PREFIX       ::=     PN_CHARS_BASE ((PN_CHARS | '.')* PN_CHARS)?
> > [166s]  PN_CHARS        ::=     PN_CHARS_U | '-' | [0-9] | #00B7 | [#0300-#036F] | [#203F-#2040]
> > [164s]  PN_CHARS_U      ::=     PN_CHARS_BASE | '_' | ':'
> >                                                      ^^^^^
> > Is that intentional?
> 
> Yes.
> http://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/index.html#sec-changelog
> 
> The other major grammar addition is the support for PREFIX (without
> the @) as seen in SPARQL.

Sure, it says: "Local part of prefix names can now include ":"".
But the way I understand the sentence this is about the local part, not
the prefix name itself. Am I missing something.

To illustrate the problem consider this prefixed name:

 abc:def:ghi

What part is the prefix, what part is the local name? Especially if the
document also contains this directives:

 @prefix abc: "http://www.example.com/first"
 @prefix abc:def: "http://www.example.com/second"

Not saying this is impossible to implement, but I expect unnecessarily
inefficient implementations, and also interoperability issues by
allowing colons in the prefix name. What's actually the use case for
permitting colons in the prefix name?

Ciao,
Mathias

Received on Tuesday, 17 July 2012 07:23:11 UTC