Yeah, I didn't mean it's a unit test as it'd already been quite heavyweight.

It's much better than I thought - with the image pulled the test takes about 5s 
compared to the previous 6-7s. I'll take this approach then, thank you for your 
advice.

> You could argue it's already an integration test. Running an integration
> test against an in-memory implementation doesn't make it a unit test (IMO).
> I think it's reasonable to run integration tests that execute against
> testcontainers in the PreCommit. The bar for PreCommit should be how
> expensive/time-consuming the test is compared to how much signal it gives
> us, not whether we classify it as a unit test.
> 
> Have you compared the runtime when running the test with testcontainers?
> 

Reply via email to