RE: [AF] relative URIs for attachments


>Thank you for drafting this, it looks to me like a good start. 
> One point 
>of concern.  I don't think it's appropriate to say that a resource is 
>"identified" by a URI reference.  As I understand Web architecture, 
>resources are identified by URIs.  URI references are used to refer to 
>such URIs, and they allow some convenience forms (notably 
>relative) that 
>require conversion (to absolute) before the referent can be identified.

I agree that we should use the term "URI" and not "URI Reference"
throughout. However, in my proposal that I just sent out [1], I
suggested a separation between section 4 (model) and 6 (implementation)
that would cause the use of relative URIs to be mentioned in section 6
rather than 4.

>I'm not sure about the fragid stuff, but I URI refs support them 
>lexically, so we need to either say how to use them, or 
>explicitly rule 
>them out.  I can live with the above.  Thanks.

I am not sure I agree. Fragment identifiers' role is described in RFC
2396, section 4 [2] and TimBL's design notes [3]. In our case, all we
need is a mechanism for identifying the URI, whether there is a fragment
identifier or not doesn't really affect our use at all.



Received on Sunday, 15 September 2002 21:18:35 UTC