Cache provides a reverse cache for any content generated by the server. It is transparent, in that it obeys all relevant HTTP caching and freshness directives and also obeys the Wub-specific -dynamic directive.
Cache is invoked prior to dispatch by Httpd in order to satisfy requests from cached content, and after Domain processing to cache any suitable responses.
Cached content may be explicitly invalidated by URL, and old content will be superceded by newly generated content.
Cache ignores requests by the client to re-generate content, reasoning that it serves the server, not the client, and the server is better placed to know what needs caching.
Caching is really wrecked by spiders and bots ... their access patterns exhibit no locality of reference, expanding the working set beyond the useful confines of a cache. Some way of detecting them needs to be worked out, so content produced for them doesn't fill the cache.
(generated by STX)