Remy Blank wrote:
> Just one clarification: the cache we are talking about cannot be defined
> as database-backed. The database is the primary data storage, and we
> cache some of that data in memory. At this point, we are not talking
> about caching e.g. generated HTML pages or anything like that (yet).

Ok, I have just realized that the cache actually *is* database-backed,
as the generation and timeout info is stored in the database.

The only purpose of this table is to signal other processes when cached
data should be invalidated. I wonder what other mechanisms could be used
for this purpose. Update the mtime of a key-specific file? Pipes?

-- Remy

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to