- From: Dave Beckett <dave.beckett@bristol.ac.uk>
- Date: Mon, 13 Jun 2005 12:09:56 +0100
- To: public-cwm-talk@w3.org
On Sun, 12 Jun 2005 20:40:39 -0400, Eric Prud'hommeaux <eric@w3.org> wrote:
> cwm, Turtle and SPARQL allow trailing '.' and ';' but not ',' . I made
> a pitch to DanC to suggest that trailing ';'s are more confusing than
> liberating, but he may have convinced me otherwise. The distinction
> between ';'s and ','s seems arbitrary.
Actually Turtle at least requires a trailing '.'. As far as I know
SPARQL requires it between triples and is optional after. Don't know
about cwm/N3.
> Are we at an optimum now? Should n3.n3 (and turtle and SPARQL) change
> to allow trailing ','s?
>
> objecttail cfg:mustBeOneSequence (
> ( )
> ( "," )
> ( "," object objecttail )
> ).
>
> Should we get rid of trailing ';'s?
I prefer not from what I hear as feedback to Turtle and see people
using in data. It is much more common to have a subject followed by
a list of pred1 obj1; pred2 obj2; ... than it is to have the same
subject and predicate followed by a list of obj1, obj2, ....
Allowing a trailing ; for the pred+obj lists is friendly to what
appears more often written by humans and also to software where it's
handy to just "printf":
subj
pred1 obj1 ;
pred2 obj2 ;
...
then end with a
.
Than worry about the special case of the last item .
> I notice that we are also under-utilizing umlauts and plus there's the
> extensive set of graphics characters near kanji that are well-utilized
> in spam.
Is that a new N3 design principle you've discovered? :)
Dave
Received on Monday, 13 June 2005 11:12:06 UTC