Diagnostics to disable "reportUnknownArgumentType" for lambdas only in complex scenarios

[Kein]

Consider the following:

my_map = dict(numValues = {'reader': lambda x: unpack_int(x)[0]}

If you have reportUnknownArgumentType set to anything above none it will report this. Problem is, as far as I know, in that particular setup, you cannot annotate lambda properly with type hints. But if you disable reportUnknownArgumentType you will lose the ability completely.

I don't think that's a fair trade and I think adding extra diagnostics option is lesser evil and less crumblesome than plaguing your code with #type: ignore all the way:

[erictraut]

You are correct that Python unfortunately doesn't provide any syntax for annotating the input parameters for a lambda, but you can provide an annotation outside of the lambda that provides sufficient context.

I can provide more details if you supply a working self-contained sample. The sample you've provided above has several errors and omissions.

[Kein]

The screenshot wasn't an example, it was intended to showcase how verbose and noisy it becomes.
Yes, I can provide an example:

import struct
import binascii

b = b'\x01\x00\x00\x00'
unpack_int = struct.Struct('<I').unpack

my_map = dict(numValues = {'reader': lambda x: unpack_int(x)[0]})

for key, value in my_map.items():
	f = value['reader']
	value['value'] = f(b[0:4])

print(my_map)

I figured you can just do:

 #type: dict[str, dict[str, Callable[[bytelike], Union[int, bool, None]]]]

but the thing is, down the line I assign new keys with different value types so now I have:

#type: dict[str, dict[str, Union[Callable[[memoryview], Union[int, bool, None]], int, bool]]]

but even this wont sole the issue when I get two keys from value that are ints but they cant be summed:

I still insist that the feature in question would be useful to avoid needless noise. Perhaps implemented in this way as described here:
#282 (comment)