- From: John Franks <john@math.nwu.edu>
- Date: Tue, 12 Aug 1997 12:53:41 -0500 (CDT)
- To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
On Tue, 12 Aug 1997, Josh wrote:
>
> I request to a 1.1 server:
> GET http://www.foo.com/fo.html HTTP/1.0
>
> So, what version is this ?:
>
> HTTP/1.N 200 OK
> Server: toaster/2.0
> Content-Length: 1024
> Cache-Control: minfree=0
>
>
If the request is to a 1.1 server the response should be HTTP/1.1...
I'll assume that is what you meant. Then if "Cache-Control: minfree=0"
is defined in HTTP/1.N the message version by my definition would be
1.N. A 1.M client being told that the message version is 1.N when M < N
is not too useful, but it knows to expect headers it does not
understand. However, the 1.N version information might be of use to a
proxy between the two.
If you meant what is the message version if (as I believe is currently
the case) "Cache-Control: minfree=0" is undefined in any version of
HTTP then I don't know. I guess a more careful definition would take
account of that "experimental" case.
>
Klaus Weide wrote:
> > > As given by your formal definition, the "[message] version"
> > > can be trivially derived from the
> > > message. It just requires tables of all headers defined by the various
> > > protocol versions, nothing else. Therefore it is totally redundant.
> > >
> >
> Well, the table will presumably evolve over time, with
> protocol revisions, extensions, what have you...
> In the proxy, you wont have the same table to generate the
> version as you do in the server.. One may
> be older than the other and may compute a different
> version..
> If the computed version number was attached to the message/entity
> there wouldnt be a need to recompute it or any ambiguity.
>
And at the very least the redundancy adds robustness.
>
> At this point, its academic. I beleive that the response
> version should be the entity/message version, it seems
> that others feel it should be the server's version.
> Currently, its the server's version
>
Agreed.
> Despite which is better, it would take a serious compelling
> breakage to change it, which isnt the case.
>
Agreed.
John Franks Dept of Math. Northwestern University
john@math.nwu.edu
Received on Tuesday, 12 August 1997 10:55:36 UTC