Tue Apr 20 16:07:32 EDT 2010

Full circle

So, with the caching structured fleshed out as 2 level: 

    txt -> articles -> (key -> response)

The only remaining problem is to make explicit quat a query is.
Currently that's a bit problematic.  It might be simplest to stick to
a very raw representation of a query, i.e. a dictionary of name.value
pairs, where names are symbols and values are strings.

To obtain an assoc list like this use:

   (url-query (request-uri req))

The `req' can be obtained from dispatch-rules, which will deconstruct
an url and pass the req as the first parameter.

So there's no problem here, a symbolic key (the entry ID) and a
dictionary of optional parameters is all that's necessary for a query.

The problem is memoization though: we need to extract keys to
construct the final (req -> response) memoization, where response is
an rv.