Publisher Limit Exceeded Behavior defaults to IGNORE while Java library defaults to BLOCK

[oryband]

The default publisher limit exceeded behavior is inconsistent between Python and Java libraries. Python "silently" skips messages while Java blocks. I'm migrating an application from Java to Python, and I was surprised the behavior to be different. How come? Both libraries use thread pools without async-io (or it's java equivalent). At the very least this inconsistency should be documented and emphasized.

• Python default settings: https://github.com/googleapis/python-pubsub/blob/265f4106f499ec5d2d01a127ba192404c1836a28/google/cloud/pubsub_v1/types.py#L130
• Java default settings: https://github.com/googleapis/java-pubsub/blob/1c79ad7336f51ffd4e177be5c2f2a7de902b47b8/google-cloud-pubsub/src/main/java/com/google/cloud/pubsub/v1/Publisher.java#L697

[mukund-ananthu]

Hi @oryband , looking at the lines you refer to in the post, I see that currently:

1. Python: limit_exceeded_behavior: LimitExceededBehavior = LimitExceededBehavior.IGNORE

2. Java:

   .setFlowControlSettings(
               FlowControlSettings.newBuilder()
                   .setLimitExceededBehavior(FlowController.LimitExceededBehavior.Ignore)
                   .build())

Where both languages have the behavior as ignore. Could you please let me know where you observe this deviation?

[mukund-ananthu]

@oryband , Closing this issue as not reproducible due to the reasons mentioned in https://github.com/googleapis/python-pubsub/issues/1055#issuecomment-1979857702. Please feel free to re-open the issue if required.