- From: Shel Kaphan <sjk@amazon.com>
- Date: Fri, 18 Aug 1995 13:56:14 -0700
- To: Koen Holtman <koen@win.tue.nl>
- Cc: fielding@beach.w3.org, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Koen Holtman writes: > Jean-Philippe Martin-Flatin: > > What do we gain by having both a > >'no-cache' and a 'private' Pragma in terms of functionality ? A 'shared > >response cache' is basically a proxy/cache, and Pragma is meaningful to > >proxies only, not to user agents. So both headers really mean "don't cache > >this response in a proxy/cache", and both let a user-agent local cache free > >to cache the response or not. > > Pragma: private would instruct a user agent not to cache the response if its > cache memory (say part of a harddisk in an MS-DOS pc in a university PC lab) > is publicly accessible. This is particularly important for user agents that > do not clear their caches at the end of the session. > > Of course, the user agent needs to be configured to know that its cache is > publicly accessible, one could have a configuration option like `cache > private responses on local disk?'. > > >Jean-Philippe > > Koen. This explanation is a little different from others floating around. The other interpretation is that Private means that no other user-agent should be allowed to access a cached copy of this document, which would presumably mean that it should not be cached in any intermediate proxy (except possibly for use by the originally requesting user-agent, but that's too hard and not very useful). --Shel
Received on Friday, 18 August 1995 14:00:56 UTC