On Thu, 15 Oct 2020 22:31:25 -0600 Carl Meyer <carl@oddbird.net> wrote:
On Thu, Oct 15, 2020 at 3:33 PM David Mertz <mertz@gnosis.cx> wrote:
So the use case needs to be:
* Function operates on large objects * Function operates on large, immutable objects * Function never takes literal or computed arguments (i.e. not `fun(big1+big2)`) * Large immutable objects are deleted selectively (OK, this is plausible) * The performance hit of clearing entire cache is not suitable
One real-world use case I've seen personally that meets all these criteria is per-request memoization of expensive work in a server application (using an `lru_cache`-like decorator on functions taking the request as an argument.) The request object plus things hanging off it can be quite large, it's often effectively immutable (at least inasmuch as the functions so decorated care), it's desirable for all cached data for a request to expire as soon as handling for the request is finished, but you wouldn't want to fully clear the caches (e.g. the application may handle multiple requests concurrently, so other requests may still be in-flight.)
A common recipe for that is simply to set a custom attribute on the request: request._myapp_cached_data = ... It's probably much faster than caching through a dict, as well. Regards Antoine.