Re: draft-06 plus last last-call changes

Roy T. Fielding wrote:
> and a paragraph to the sections on scheme and absolute-URI to
> indicate that scheme specifications must define a syntax that
> is a subset of absolute-URI and thus don't have to redefine
> fragment for every scheme spec
> <>.

I think that between this and the final paragraph of section 3.1 (which I was 
overlooking), the question in my last email -- about what leeway a scheme has 
in overriding the generic syntax, what jurisdiction it has over URIs not 
conforming to its restrictions, and what should be done with such URIs -- has 
been adequately addressed.

In both places, by "subset", I assume you mean that the scheme can define any 
syntax (not necessarily in terms of rfc2396bis rules) such that a URI 
conforming to that syntax would not fail to also conform to the absolute-URI 
syntax rule. I kinda wish it were stated more like that, but I suppose it's 
good enough as "subset".

Anyway, given this bit, and the advice in sec. 3.1, I feel the RFC 1738 scheme 
definition replacement specs should all be quite clear about whether and how 
to interpret a URI that has extraneous components or that otherwise does not 
conform to the scheme's specific syntax.


Received on Sunday, 26 September 2004 00:09:35 UTC