fix: include_feedback breaks when > 10 results AND no expand fields listed
#2541
Conversation
|
Preview this PR with FeatureBee: https://beta.wandb.ai/?betaVersion=874e2383126a421baa23292d16a4e8f1a9eba727 |
| @@ -2664,6 +2664,42 @@ def return_nested_object(nested_obj: NestedObject): | |||
| assert call_result.output == nested_ref.uri() | |||
|
|
|||
|
|
|||
| @pytest.mark.parametrize("batch_size", [1, 11, 101, 120]) | |||
There was a problem hiding this comment.
this tests batch of 10, 100, and 500. 120 because what if theres an off by one error and we are just getting lucky 🧠
There was a problem hiding this comment.
You should add a comment -- these look like magic numbers otherwise. I would make the 10, 100, 500 consts and re-use them here with that comment
There was a problem hiding this comment.
Because of how the tests insert calls synchronously generating 500 calls actually gets pretty expensive, like 60+ seconds expensive. 120 completed in 15 seconds ish. I will add a comment
There was a problem hiding this comment.
yeah you dont have to do 500, just comments RE: what these numbers mean
| # double batch size up to what refs_read_batch can handle | ||
| batch_size = min(max_size, batch_size * 2) | ||
| # *** Dynamic increase from 10 to 500 *** | ||
| batch_size = min(500, batch_size * 10) |
There was a problem hiding this comment.
Let's move this to a top level constant
Simplify dynamic batch size calculation in calls stream to be
10then100then a max of500