> I was thinking that it could be implemented as a keyword or comment in a query.
> Such as:
>
> select * from table where column = 'foo' cacheable
> or
> select * from table where column = 'bar' /* cacheable */
> Having caching within PostgreSQL will be good for data integrity. Application
> caches can't tell when an update/delete/insert happens, they often have to use
> a time-out mechanism.
>
> OK, let me have it, tell me how terrible an idea this is. tell me how wrong I
> am.
I don't think it's a bad idea, but a cache that takes a query string (or
subquery string) and looks for a match based on that is flawed without
special consideration to non-cacheable functions and constructs
(CURRENT_USER, things that depend on timezone, things that depend on
datestyle). We'd also need to work out an appropriate mechanism to deal
with cache invalidation and adding things to the cache.