accessing cached objects direct from the cache

From: George Michaelson <>
Date: Thu, 19 Mar 1998 12:19:15 +1000 (EST)

We've come up with a scheme to re-use the objects in the cache explicitly
from the cache. It depends on exposing the hash function from the URL
so a client can put in a GET cache_object: request in some efficient
manner to let the cache send back the specific object.

This has some 'social engineering' consequences, like bypassing the
cookie and other access control on the real object. None-the-less,
the data is in the cache, its known to be in the cache, and it could
be bloody useful to be able to 'see' it direct from the cache.

        Does anybody else think it would be 'useful' to have a way
        to request a cache_object: form of URL which allowed the cache
        to return the content of another specific URL: form known to be
        in the cache?

the cachemgr.c code seems to have 3/4 of whats needed. the unimplemented
REMOVE function for instance, takes a url as an argument. I can see that
it would work to display that url if in the cache. The idea to use the
hash on the name is to bypass having to compute that inside the cache but
if the cost saving is too low to be worth it, then it would be simple to
just get the object.

Why not just GET <url> I hear you ask? Because specific remote sites are
now tagging their data so that the IMS request to the thing fails because
associated headers like referrer are missing. I'm split over the merits
of bypassing this kind of thing, but rather than pre-empt the issue I'd
like to see some feedback from other cache-droids.

Received on Wed Mar 18 1998 - 18:30:35 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:39:25 MST