W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 1997

Re: more on digest (was: Unidentified subject!)

From: Roy T. Fielding <fielding@kiwi.ics.uci.edu>
Date: Wed, 17 Dec 1997 11:11:19 -0800
To: John Franks <john@math.nwu.edu>
Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Message-Id: <9712171112.aa07459@paris.ics.uci.edu>
>The only reason this came up at this point was that because a hash
>of the Date, L-M and Expires headers can be part of the response
>there could be a problem for servers with no clock if a proxy added
>a Date header.  There is a simple answer to this which is that
>proxies should not be allowed to add or change Date, L-M or Expires
>headers.  There are no known implementations which do so and no one
>has suggested any reason it is necessary to do so.

An HTTP/1.1 cache is required to change Date and Expires upon receipt
of a 304 response containing updated values for those fields.  This
does impact non-shared caches, so you will need to add something to the
effect of the digest should be removed if those fields are updated.

The Apache proxy canonicalizes the response field-values of Date,
Last-Modified, and Expires to the required HTTP-date format.  I have
no idea what effect this would have on entity-digest.  If it caches
the response, the cache will add Date and Content-Length if they are
missing, but it won't normally cache the response if the request
included Authorization (this would not be the case if we ever
developed a personal, non-shared proxy).

....Roy
Received on Wednesday, 17 December 1997 11:34:19 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:33:05 EDT