Hi -- I'm interested in eliminating type annotation errors on flyte workflow files as reported by

pyright

or

mypy

. We recently updated to

flytekit==1.10.2

(from

1.7.0

). I first removed the decorators

@dataclass_json

based on PR-1801, and that does eliminate many of the false positive reported by the static type checking tools. But I'm still seeing a lot of errors like this:

error: Expression of type "Tuple[Promise] | Promise | VoidPromise | Tuple[Unknown, ...] | None" cannot be assigned to declared type "TrainDatasetParameters"
error: "VoidPromise" is not iterable
error: Expression with type "Tuple[Promise] | Promise | VoidPromise | Tuple[Unknown, ...] | None" cannot be assigned to target tuple
error: Operator ">>" not supported for types "Tuple[Promise] | Promise | VoidPromise | Tuple[Unknown, ...] | None" and "DataFrame"

I am also getting a few new type check errors like this:

Argument of type "PythonFunctionWorkflow | ((...) -> bool)" cannot be assigned to parameter "workflow" of type "WorkflowBase" in function "get_or_create"

But I think these might be real.

okay. would you mind creating an issue?

I added some information to https://github.com/flyteorg/flyte/issues/3682 I think that existing issue is a good summary of the situation. I added an even simpler example showing the problem.