I have a strange request for cache behaviour and would like to know if it's doable without too much coding.
Having cca 70% requests served to anonymous users (which all return same for same URL) and 30% requests served to a class of users identified by a presence of one or two cookies (everybody gets different page) I would like to have that 70% requests processed & served by cache and those 30% forwarded directly to a backend web server (webserver's respond in this case should no affect cached version of page in any way ... except maybe gziping output when client supports it) -- for those 30% cache server should behave as not being there (proxy would be set in a http acceleration mode and no direct with the outside world would be permited for web server). If this feature is unavailable in current squid ... would it's implementation be at least semi maintainable or would it change codebase way too much? Regards, Primoz