`typing.IntoSchema` too narrow

[MarcoGorelli]

This runs fine, as expected:

print(nw.from_dict({'a': [1,2,3]}, schema={'a': nw.Int64}, backend='pyarrow'))

but the currently typing wants us to write

print(nw.from_dict({'a': [1,2,3]}, schema={'a': nw.Int64()}, backend='pyarrow'))

[dangotbanned]

but the currently typing wants us to write

That was because I read the docs to the letter for that case, but not for #3176 🤦‍♂️

narwhals/narwhals/schema.py

Lines 45 to 53 in bc6a77f

	class Schema(OrderedDict[str, "DType"]):
	"""Ordered mapping of column names to their data type.
	Arguments:
	schema: The schema definition given by column names and their associated
	*instantiated* Narwhals data type. Accepts a mapping or an iterable of tuples.
	Examples:
	Define a schema by passing *instantiated* data types.

I was thinking about this the other day actually, it can probably use IntoDType?

[MarcoGorelli]

sounds good!

I was thinking about this the other day actually

did you already have something in the works, or shall i make a pr?

[dangotbanned]

did you already have something in the works

Nope, just on my mind with how many times I (originally) got the warning in these tests 😂

• https://github.com/narwhals-dev/narwhals/blob/d09aa748fa0090970cdb9f76d408010e225ac90f/tests/plan/selectors_test.py

or shall i make a pr?

Absolutely go for it 🙌

[dangotbanned]

Oh actually one thing!

Schema might need to do the instantiation internally.

Otherwise some methods that return a DType will be able to return type[DType]

Related?

• polars.Schema check_dtypes parameter is undocumented pola-rs/polars#24784