Re: Quetions about cwm/N3

At 03:35 PM 5/7/02 +0100, Sean B. Palmer wrote:
> > (1) what is the keyword 'this' meant to mean?  I think I
> > read somewhere that it's the enclosing formula.  In the
> > absence of {...} constructs, that could be replaced by a
> > URIref '<#>', I think.
>It does identify the encosing formula (TimBL explained it as something like
>"<> log:semantics this ."), but you can't use the URI-ref <#> for that,
>since the fragment identifer syntax of XML RDF does not currently specify
>that the empty fragment identifies the enclosing formula (it could be made
>to do so, I suppose). Note that CWM uses <#_formula> to identify the root
>formula: a nasty hack. FWIW, my latest API treats it as a completely blank
>node (labelling it with Python's "None").

Aha, that's clearer.  I've just tweaked my parser to handle that.  (But the 
result is very different than cwm because I treat nested formulae very 
differently - I store them as distinct collections of statements;  the 
#_formula hack doesn't arise because I always have a node to contain any 
formula -- even the top-level formula.  This is following the way I propose 
to handle contexts -

> > (2) is the character '-' intended to be allowed in QNames?  I
> > thought not, but I've come across some cwm output that contains
> > this character.
>Try: TimBL reserved it in the
>DesignIssues document, but many implementations still support it (note the
>great "-" to "_" mapping solution in the N3 DesignIssues document).

Cool.  Thanks.  FWIW, my parser (independently) follows your recommendation 
(except that it's temporarily hacked to allow '-' because I wanted to 
process some output from cwm).

I don't think the liberal/conservative argument can really be applied 
here:  the form of output one generates will generally depend on the input 
one is given (unless one gets into rewriting names, which sounds to me like 
a great opportunity for confusion and non-interworking).


Graham Klyne

Received on Tuesday, 7 May 2002 12:11:32 UTC