- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Mon, 08 Apr 96 17:45:27 MDT
- To: Larry Masinter <masinter@parc.xerox.com>
- Cc: http-caching@pa.dec.com
I just had a lengthy conversation with a couple of folks developing
web browsers for PDAs, where they expect long periods of disconnected
operation and will want to cache pages until their devices come back
in IR range. This application seems like a perfectly reasonable one
where the cache is going to ignore any 'max-stale' requests, and I'd
like to make sure that the wording in the spec makes their application
continue to be legal, at least with some caveats.
The spec specifically allows a cache to disobey "max-stale" directives
(on requests or responses), but it must attach a Warning to any
stale response.
The spec specifically prohibits a cache from disobeying "must-revalidate"
directives, period.
In order to meet Koen's requirements for reliable cookie support,
the spec has to have a MUST someplace in it.
If your PDA people want to implement a *history* mechanism, that
doesn't have to obey the rules for caches. If they treat it as
a cache, it has to obey the rules.
>I.e., you need an unavoidable mechanism to require at least conditional
>validations on every access.
but this really isn't possible. At best, the implementor of the PDA
application can warn the user that the information is out of date.
The PDA application would say "failed to contact server".
-Jeff
Received on Tuesday, 9 April 1996 01:05:28 UTC