Re: HTTP/1.1 headers

G'day Miguel!

Miguel A.L. Paraz wrote:
> Subject: HTTP/1.1 headers
> Ed Knowles wrote:
> > - Date: isn't used (that I know of)
>
> it's mandatory, though.

Yes, but that doesn't make it right :)) Why not make VALID Expires: mandatory?
That would make much more sense.

> > - Cache-Control:, Expires:, Last-Modified:, ETag: all can have some
> > influence on if the object is fresh or not, with only one needed, the best
> > being a toss up between Expires:/Cache-Control: (same thing, though if you
> > want to split hairs, one is shorter than the other) and ETag: (though this
> > requires a connection to the site) so probably make is second.
>
> Last-Modified is used by the cache for "guessing" the possible lifetime.

Yes, but no need to guess if you have a valid Expires:

> As for Expires/Cache-Control, which would be more authoritative,
> I have a flag in my mod_expires.c Apache module that allows you to turn
> off Expires if Cache-Control is present; I left it on in my site "just in
> case" there would be some other caches (not Squid) who would

Draft-ietf-http-v11-spec-07.txt says a Cache-Control: max-age=nn overrides an
Expires: ... seems odd to keep both.

After a little more reading, ETag: should superscede an IMS request, with the
use of an If-Match: request. ETag: can also be used as a check with a PUT
request. Looking at the NLANR caches, ETag: isn't as popular as Expires: (which
isn't that popular anyways) but they are being used.

There is also an Age: header for a cache to play with.

> Finally, I think our discussion would be great for the mailing list so
> the other folks can join in; let's bring it back there?

I think we are stuck with the current headers, but if you know a way to
consolidate, please let me know :)

Later
Ed

-- 
Ed Knowles aka Jasper				   Phone : +61 2 9385 4962
E-mail: ed@fatboy.geog.unsw.edu.au	           Fax   : +61 2 9313 7878
            What I lack in morals I make up for in principles.