- From: Lou Montulli <montulli@mozilla.com>
- Date: Mon, 13 Nov 1995 16:25:50 -0800
- To: Simon Spero <ses@tipper.oit.unc.edu>
- Cc: Larry Masinter <masinter@parc.xerox.com>, ietf-lists@proper.com, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Simon Spero wrote:
>
> On Mon, 13 Nov 1995, Lou Montulli wrote:
>
> > Simon Spero wrote:
> >
> > We don't need a hack here. Using a 205 response to signify a
> > partial document is being returned seems far better than
> > the "no-cache" nonsense. The 205 response is also necessary
> > for the client to tell the difference between a full document
> > and a partial document response.
>
> Lou- the 205 response is not part of http/1.0 as described, and this can
> lead to some caching confusion; if a 1.1 client is talking through a 1.0
> proxy to a 1.1 client, and a partial fetch is done, it's possible to the
> 1.0 proxy to keep a copy of the partial contents, yet not be aware that
> the contents are bogus- the next client to do a fetch could end up with
> partial data, yet not be aware of it.
I see, so we are back full circle to the reason why byte ranges were
originally proposed as part of the URL.
But your concerns are not valid. Neither of the two most used proxy
servers (CERN and Netscape) will cache objects that are not returned
with a 200 status. If there are any other proxy servers that do
cache non-200 requests then they are clearly in violation of the
HTTP spec.
:lou
--
Lou Montulli http://www.netscape.com/people/montulli/
Netscape Communications Corp.
Received on Monday, 13 November 1995 16:34:00 UTC