Re: TB16 Re: Comments on arch doc draft

Tim Berners-Lee wrote:

> I think we should establish at the strong SHOULD level that
> relative URIs should/must be allowed in a syntax wherever 
> absolute ones are.
> 
> The only example I can think of where they are not
> allowed (xmlns) was shown to be a problem at our Hawaii
> tag face-face, when most of the GET URI for a query was
> a namespace URI which could have been dramatically
> shortened if made relative to the URI of the query itself!

I think that the issues surrounding namespaces and relative URIs were 
explored exhaustively, with the investment of several thousand email 
messages, back in 2000; the eventual finding was that relative URI 
referencess should not be used for XML namespace names.  If we are going 
to re-open this issue in the TAG context with a view to reversing it, 
that should be a deliberate policy decision and not one we take lightly.

I don't think the trick we outlined in the strawman for url-encoding 
SOAP requests changes the terms of the debate at all; it packs two 
references into the same URI, the "relative" piece could not exist 
outside of the context of this particular URI, and it's not clear that 
the relative piece was actually an XML namespace name anyhow, it was a 
service ID. -Tim

Received on Tuesday, 2 July 2002 12:29:23 UTC