wandgitlabbot
commented
3 years ago In GitLab, by Daniel Oosterwijk on 2021-01-22
Turns out there wasn't that much to fix about the semantics, but I've refactored it a bit anyway. I'll copypaste the description of the commit here:
- Memcached backend now uses Kryo for serialization instead of whatever scalacache used to give us by default for Option[Any].
- Improved documentation and abstractions in PostgresConnection.
- Caching now uses T instead of Option[T], giving us better compile-time guarantees.
Of course, there's no direct tests for Caching, so let's fix that in a new issue.
In GitLab, by Daniel Oosterwijk on 2020-10-14
PostgresConnection has a strange relationship with Caching. Not all of its functions use caching, seemingly arbitrarily, but some of the places that use the connection then go ahead and cache the results that the connection doesn't cache! This really should be ironed out, and we might want to look at improving the TTL interface and how to request fresh results.