Re: want to understand how back-ldap works...

[Pierangelo Masarati <masarati@aero.polimi.it>]

> [copied to -devel, it seems to be at least of historical interest and the
> caching is certainly a -devel topic]
> 
> If all you want to do is cache search results, certainly it can be done and
> you only need to hook into back-ldap/search.c. I suppose you could copy the
> cache management approach that back-ldbm uses.

I note one major issue that prevents a "naive" caching (I think
we already discussed it) is that entries are usually retrieved
partially by the proxy for two/three reasons:
	a) a search may request only a subset of the attributes;
	b) some attributes can be hidden due to ACLs;
	c) some attributes may be dynamically generated
Issue (c) falls in cache refreshing; issue (a) may be circumvented
by requesting ALL the attributes for caching purposes but returning
only the requested ones; issue (b) can be overcome only caching
on a per-user base; this means the space of the cache might grow to
(n. entries) x (n. users).

Pierangelo.