The test assertions raised by ModelTaggingCheck are reversed from what you'd expect.
I can kind of see what I was thinking: the actual side of the comparison is what you've got in the class, while the expected is what we generated from actually looking at the flows in the model, so it sort of makes sense.
In practice it's easy to get an assertion failure and automatically think "OK, I'll copy the actual result into the class and re-run", and just get the same error again.
It's obviously frustrating to have to stop and think when that achieves nothing.
therealryan
commented
1 month ago
The test assertions raised by ModelTaggingCheck are reversed from what you'd expect. I can kind of see what I was thinking: the actual side of the comparison is what you've got in the class, while the expected is what we generated from actually looking at the flows in the model, so it sort of makes sense.
In practice it's easy to get an assertion failure and automatically think "OK, I'll copy the actual result into the class and re-run", and just get the same error again. It's obviously frustrating to have to stop and think when that achieves nothing.
We should flip the polarity of that assertion.