You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There's a lot of weird undocumented stuff in impl/memory which was added after discovering that behavior in production. It would be great to be able to pull each weird thing out into a tiny 'documentation app' which uses the stock go SDK to form a minimal test case which can be uploaded to an appspot instance to demonstrate the bug/limitation. This would also help us remove these weird things when they eventually get fixed (since the goal of impl/memory is to provide an as-accurate-as-possible replication of the production behavior as possible).
Some examples of this include:
limits on # entities modified in a transaction
discrepancy between unset namespace and "" namespace in some APIs (like datastore_v3.RunQuery)
oddball combinations of Query parameters which don't work/behave in surprising ways
limits on # simultaneously modifiable entities in e.g. PutMulti
maximum RPC size limits
etc.
Note that impl/memory currently doesn't implement some of these, as it doesn't actually implement the RPC middleman. If we fix #23, some of these will become more pertinent.
The text was updated successfully, but these errors were encountered:
@danjacques
There's a lot of weird undocumented stuff in impl/memory which was added after discovering that behavior in production. It would be great to be able to pull each weird thing out into a tiny 'documentation app' which uses the stock go SDK to form a minimal test case which can be uploaded to an appspot instance to demonstrate the bug/limitation. This would also help us remove these weird things when they eventually get fixed (since the goal of impl/memory is to provide an as-accurate-as-possible replication of the production behavior as possible).
Some examples of this include:
""
namespace in some APIs (like datastore_v3.RunQuery)Note that impl/memory currently doesn't implement some of these, as it doesn't actually implement the RPC middleman. If we fix #23, some of these will become more pertinent.
The text was updated successfully, but these errors were encountered: