mark nottingham

httpRange-14

Tuesday, 29 July 2003

HTTP

Mark Baker is the latest in a series to weigh in on the TAG issue regarding what a HTTP URI can identify.

I haven’t followed the debate closely, but it appears that the arguments haven’t changed substantively. Another way of stating this issue is asking whether a URI scheme fundamentally constrains the types of resources it identifies.

If you look at it that way, the scheme component of a URI is fairly special; it’s a declaration of support for a particular interface to the resource (in the case of http URIs, the protocol described by RFC2616).

Of course, this doesn’t constrain the nature of the resource itself; you could have an FTP interface to it as well, and/or a transporter beam interface (gotta future-proof this stuff, you know).

At the end of the day, however, those different interfaces to the same conceptual resource will have different URIs, because they have different schemes. Which raises the question - do URIs identify resources or do the really just identify discrete interfaces into them?


One Comment

Yves said:

I advocated for the concept of “views” of a resource a long time ago. Exporting a resource using different protocols is very commonly done (file and HTTP). You use another wording (interface to resources instead of views), but the distinction is the same, interface/view or plain resource. With the experience of running code, I am all on the “interface/view” side.

Saturday, August 9 2003 at 7:43 AM