Implement Global Cache (memcache) #148
Conversation
googlebot
added
the
cla: yes
src/google/cloud/ndb/client.py
Outdated
| @@ -142,8 +148,23 @@ def context(self, cache_policy=None): | |||
| cache_policy (Optional[Callable[[key.Key], bool]]): The | |||
| cache policy to use in this context. See: | |||
| :meth:`~google.cloud.ndb.context.Context.set_cache_policy`. | |||
| global_cache (Optional[_cache.GlobalCache]): | |||
There was a problem hiding this comment.
Might need to put the whole path here to appease sphinx fairly inconsistent reference finder.
There was a problem hiding this comment.
I think it's actually the fact that _cache isn't autodoc'ed yet. I plant to make a docs pass before merging this.
|
Looks like the global cache access is happening in the datastore_api module, instead of in the user-facing functions like get and put. Is there a particular reason you made that decision? |
|
Yes. The global cache stores serialized Datastore protocol buffer representations of entities. The _datastore_api layer stores and retrieves protocol buffer representations, so it made sense (to me) to put the global cache there where it's speaking the same language. This allows data to flow from low level Datastore protocol buffers to high level NDB entities in a single, straightforward pipeline. |
|
HI @cguardia , I've rebased and added one more commit, if you want to take another look before I merge. |
Does not include:
Concrete implementation of
GlobalCachethat uses Redis. (Should only take a day or so.)use_datastoreimplementation.Makes use of @takashi8 's surprise #130 contribution.
I still need to work on documentation, but this should be working now. This is extensive/tricky enough, that I'm likely to have overlooked something. Happy to have this one scrutinized fairly closely.