Re: Claification requested in Host:

> Yep, which is why I said we don't need to specify it in HTTP. All of
> the above is referenced as the definition of a legal Internet host,
> and therefore all of the concerns you expressed are already covered
> by RFC 2068 via these references.  I don't see why the HTTP spec should
> create additional requirements to address those concerns.
> 
> .....Roy

OK, but then the 1.1 spec should require an error when
an incomplete name is received, because it violates the spec.:

> From fielding@kiwi.ICS.UCI.EDU Wed Feb 19 12:25:49 1997
> 
> >How does a proxy handle this?
> >This seems like something that should be nailed down in the 1.1 spec.
> >
> >Consider:
> >
> >	client xxx.foo.com requests 
> >		url http://www/file.html 
> >		from proxy proxy.com
> >
> >	proxy.com gets request, and looks up what?
> >
> >		www.proxy.com (www in the local context, potentialy)
> 
> Yes, or just respond with an error (always an option).

*MUST* be an error.
(and this is where the 1.1 spec should have something to say)

> >The 1.1 spec should require complete names,
> 
> No. The 1.1 spec should not require that a client never do something
> which does work under some circumstances, just because it doesn't work
> in all circumstances.  If the client requests on an ambiguous URL,
> it will get an ambiguous response behavior, which is as it should be.

Incorrect - as noted, it already does.

> In any case, this is a question of how to interpret the DNS hostname
> within an "http" URL, and applies equally to any URL scheme that uses DNS.

This is completely true, however.

Joe
----------------------------------------------------------------------
Joe Touch - touch@isi.edu		    http://www.isi.edu/~touch/
ISI / Project Leader, ATOMIC-2, LSAM       http://www.isi.edu/atomic2/
USC / Research Assistant Prof.                http://www.isi.edu/lsam/

Received on Wednesday, 19 February 1997 15:50:21 UTC