[airbyte-cdk] entrypoint wrapper should use per-stream state not legacy format

[brianjlai]

After deprecating legacy state from the CDK, I didn't update the CDK mock server test wrapper to refer to the per-stream state format instead of deprecated legacy data field. This is blocking updating some of our concurrent connectors that have mocker server tests since we're asserting on a field that is no longer populated.

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Ignored Deployment

Name	Status	Preview	Updated (UTC)
airbyte-docs	⬜️ Ignored (Inspect)		Mar 11, 2024 10:12pm

[erohmensing]

Seems sane! Can you point to where things start failing (since they didn't start failing when the code was released?)

Is it that we are just now getting all the old-state-emitters up to date, and if so, are there any that we can't feasibly update in time that will start failing as a result? do we need to be able to handle both in an interim state?

(mostly questions from a learning standpoint and to make sure we've covered our bases - no real concerns as this is test code, easily revertible and you know way more than I do about it)

[brianjlai]

@erohmensing

Can you point to where things start failing (since they didn't start failing when the code was released?)

Sure, so taking this Stripe mock server test as an example. We assert on that map which will 100% fail because we try to get the data field instead of the stream field.

Is it that we are just now getting all the old-state-emitters up to date, and if so, are there any that we can't feasibly update in time that will start failing as a result? do we need to be able to handle both in an interim state?

Not that I'm aware of. I just opted to get rid of legacy because it wasn't being used anymore. We shouldn't need to worry about interim state because the platform/destinations are operating on per-stream. The only impact was connectors that used mock-server tests because the bug was in the test code.