Bertrand Delacretaz wrote > On Wed, Apr 28, 2010 at 5:00 PM, Felix Meschberger <[email protected]> wrote: >> On Wed, Apr 28, 2010 at 12:35 PM, Vidar Ramdal <[email protected]> wrote: >>> ... The question is how useful such a filter would be if only the >>> last-modified date of the requested resource is used. >> >> That *is* in fact a valid concern, which my proposal does not account >> for yet. I think this also aligns with what Bertrand has in mind with >> extensibility of the basic mechanism.... > > Cocoon has a similar problem when caching xml processing pipelines. > > That's solved there by having the pipeline compute a cache key (etag > in our case?) where each pipeline element contributes part of the > cache key, from left to right. If the cache key is different once you > reach the end of the pipeline, that means the content must not share > the same cache location. > > We could imagine something similar when several resources are > aggregated to produce a response. > What about an event based cache?
Just a rough idea... During processing we know which resources are used to render the page, this info is stored in the cache. We already have events when resources change. The cache listens for the events and invalidates accordingly. Carsten -- Carsten Ziegeler [email protected]
