W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1999

RE: Fragments in "Location" field

From: Larry Masinter <masinter@parc.xerox.com>
Date: Mon, 9 Aug 1999 08:40:09 PDT
To: "Scott Lawrence" <lawrence@agranat.com>, <http-wg@hplb.hpl.hp.com>
Message-ID: <001901bee27d$75f914c0$15d0000d@copper.parc.xerox.com>
> > > The HTTP syntax should be
> > >
> > >    Location       = "Location" ":" absoluteURI [ "#" fragment ]
> >
> > I suppose this belongs in the errata, although we might need to explain
> > the circumstances under which a fragment identifier is appropriate.
> 
> I'll update the errata page.
> 
> When would it not be appropriate? I can't think of a case... but then it's
> Monday morning.

Well, let's see. 

- I don't think it's appropriate to have a fragment
  with a 201 Created response, e.g., to a POST.

- I don't know what it would mean to return a fragment with a 300
Multiple Choices, since the choice decision is intended to be made
on resource characteristics and not fragment characteristics (?).

- I think we're mainly concerned with 301/302/303/307, but I don't think
it makes sense to POST to a URI with a fragment.

- I don't think it's appropriate to return a fragment with
 305 Use Proxy.

In addition, we should specify the behavior in the case where
there was a fragment with the original URI, e.g.,
  http://host1.com/resource1#fragment1
where /resource1 redirects to
  http://host2.com/resource2#fragment2
is 'fragment1' discarded? Do you find fragment2 and then
find fragment1 within it? We don't have fragment combination rules.

Larry
Received on Monday, 9 August 1999 17:12:40 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:33:32 EDT