RE: consequences of hashes controlling cache refresh

There are other ways to get 'location' from a hash than
by also providing a URI. For example, in HTTP, a new response
header could request the sender f an accept-features to
supply an expansion of a hash.

In some circumstances, the URI for expanding multiple hashes
in the various feature expressions could be constructed
using a hash-expansion base, where the URL for a given hash
could be constructed by appending the hash to the base.

In both of these situations, there might not be a 
one-to-one relationship between "hash" and supplied URL; in
the first case, there is no URL corresponding to the
source of the data for the hash, and in the second case,
only one URL element is supplied, and the URLs constructed.

Larry
-- 
http://www.parc.xerox.com/masinter