Store users locally to improve performance when querying labs/nodes#199
Store users locally to improve performance when querying labs/nodes#199
Conversation
| self._users_by_id = {user["id"]: user for user in user_list} | ||
| self._last_sync_users_time = time.time() | ||
|
|
||
| def sync_users_if_outdated(self) -> None: |
There was a problem hiding this comment.
ok. I disliked this design before, and I dislike it now as well. I feel like this is only warranted in cases where sync method requires some additional stuff to do, if its a simple TTL mechanism, I feel we would be better off with memoization/cache.
There was a problem hiding this comment.
I think using TTLCache (from cachetools) or similar memoization object should be sufficient... we can create wrappers and separate TTLCache for each feature that would require its own cache and set TTL to 1+ seconds
user_cache = cachetools.TTLCache(maxsize=1, ttl=1)
...
class user_management:
@cached(cache=user_cache)
def users(self):
url = self._url_for("users")
return self._session.get(url).json()
def sync_users(self):
user_cache.clear()
its just an idea... implementing same method for dozen of managers/features seems very redundant
There was a problem hiding this comment.
I don't like it either but it works and I just went with what is already proven to work. This could be refactored but it should be done across the code and imo it's not a priority.
2 participants
Fixes #153