Re: BagID issue link

>Ha!  I think PatrickS was right :)
>
>The test case Jeremy presents can be made deterministic.
>
>The problem is that the XML spec says that the order of the 
>attributes is not significant.  But we can define an ordering to 
>determine in what order to add the statements to the bag, e.g. we 
>could say sort the attributes by their qnames and insert in that 
>order.  It would take a little more than that to precisely define 
>the ordering, but you get the idea.  This is sufficient to order 
>them since they attribute names must be unique.
>
>Two questions:
>
>   o does it work?

Maybe, but it might not, and it looks like a quagmire to me. Eg we 
will need to define that ordering very precisely in a way that nobody 
can argue about (...qnames???)

>
>   o is it worth it, or do we just accept the non-determinism.

Id say accept the nondeterminism. The worst that can happen is that 
poor dumb RDF has two bags that it gives different names to, but a 
more bag-savvy engine can see are really the same. But RDF is always 
in this condition; two urirefs *might* co-denote in a way that is 
invisible to RDF, that's always a possibility.

>Anyway, it seems like syntax issue, not a MT one.

Agreed. I think this is only an issue at all because of a lingering 
feeling that RDF ought to be something like a programming language. 
THe moral is, get used to the fact that it isn't a programming 
language.

Pat

-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Monday, 11 February 2002 11:37:08 UTC