Re: Review of RIF in RDF

On Sun, 2010-09-26 at 11:51 +0100, Polleres, Axel wrote: 
> The "may" sounds ok for me. Would you mind checking the current doc on the wiki and suggest concrete rewording (I will be able to check tomorrow still, hopefully). 

Done. I wasn't comfortable with the way the "<>" case was described so
I've changed that, plus added in the additional remark. It seemed easier
to edit the document than explain the suggested edits - sorry to break
protocol but we don't have much time left and that was the easier route
for me. Feel free to revert. 

> > groups in the source documents 
> > into a single group and since order
> > doesn't matter then there is no 
> > ambiguity.
> 
> Ok, but don't we have problems with order-prservation for dialects that would need order anyways already by  the current reverse mapping?

The mapping and thus the reverse mapping preserve order within a rule
set (because it is uses lists) so if the graph only encodes a single RIF
document then life is easy.  If the graph encodes multiple RIF
documents, e.g. by virtual of RDF graph merge, then we have no
definition of how to do the rule set merge. We could specify one, say
the order is arbitrary and put that specification in the future full
version of section 6. By adding the "May" remark in we keep this option
open.

On Sat, 2010-09-25 at 18:57 +0200, Axel Polleres wrote: 

> Particularly, I also added another remark:
> 
> @@@ What about several RIF documents encoded in one RDF graph (by merge of RDF encoded RDF graphs?
> 
> Can't XTr simply return a set of RIF documents?

It could, though we would still have to say how a processor is supposed
to handle a set of documents in a place where it was expecting one
document. 

> p.s.: my prblem is that I am traveling almost the whole Monday and Tuesday
> If we can clarify anything by Monday (I might have short internet access in the morning), let me know.
> One question: I understand that the RIF WG wants to close this, but SPARQL needs this document desparately.
> Could the RIF WG delegate this document to be finished within the SPARQL WG (if Dave could 
> commit to give advice/review on any remaining issues? 

I could certainly advise/review on remaining issues but I'm not sure handing over this document to SPARQL WG is an good solution procedurally. After all SPARQL WG would then have to review and vote on it and that would open up a whole new can of worms that is better avoided I think.

Dave

Received on Sunday, 26 September 2010 13:52:06 UTC