W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1998

caching & revalidation, followup -- a question to proxy server authors

From: Daniel Hellerstein <danielh@mailbox.econ.ag.gov>
Date: Tue, 05 May 1998 17:25:56 -0400
Message-Id: <s54f4be1.049@MAILBOX.ECON.AG.GOV>
To: http-wg@cuckoo.hpl.hp.com
Yesterday I inquired about how an origin server may inform
downstream caches (say, on proxy servers) that a given
response may change suddenly -- implying that a cache
may store the content, but should always revalidate before using.

rom the responses I've recieved (thanks Jeff), it appears that a 
   cache-control: max-age=0,must-revalidate
or 
  cache-control: nocache
can have this effect. However,  this is subject to implementation,
with perhaps the latter hinting that "it's not worth storing this response". 

My question is to current or potential authors of proxy servers
(or other caches).:

 Should a conscientious web server use a particular response
to hint that 
  a) this is subject to change -- so store it but always revalidate
as opposed to 
 b) don't bother storing this -- it's request specific?

For example, would a max-age=1,must-revalidate be a reasonable
way of signaling this (with a tiny chance of sending a no longer
valid stored response)?   Or should origin servers not worry -- let the
caches manage their disk space as they see fit?
Received on Tuesday, 5 May 1998 14:30:22 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:33:18 EDT