Re: relative uri references

On 24 Nov 2008, at 16:55, Boris Motik wrote:
[snip]
> The question whether we need a base resolution mechanism is a  
> completely unrelated story. The use case of being able to move the
> files around is compelling, and we can/should discuss it  
> independently from the state of the current spec.

Reflecting on it, I don't find it all that compelling. Import  
statements are all at the top (unlike, say, img srcs). The technique  
doesn't work now even if you stick to RDF processing and it's not  
been exactly a hugely requested feature. It can be handled pretty  
easily with implementation specific extensions. There are loads of  
serialization oriented stuff one elides at the moment anyway  
(including comments, order of axioms, duplicates, namespace prefixes,  
entities, etc.).

I'm not saying it's a pointless use case by any means, just that it's  
not all that common in my experience, there are loads of work  
arounds, and it actually complicates things in model.

Cheers,
Bijan.

Received on Monday, 24 November 2008 17:29:30 UTC