Type hints, or lack thereof, causing massive LSP errors. Why?

[HackNSnack]

Hey, all. I wanted to ask a question about the Python KFP SDK, where I copied an example directly from PyPI, e.g.:

from kfp import dsl
import kfp


@dsl.component
def add(a: float, b: float) -> float:
    '''Calculates sum of two arguments'''
    return a + b


@dsl.pipeline(
    name='Addition pipeline',
    description='An example pipeline that performs addition calculations.')
def add_pipeline(
    a: float = 1.0,
    b: float = 7.0,
):
    first_add_task = add(a=a, b=4.0)
    second_add_task = add(a=first_add_task.output, b=b)


client = kfp.Client(host='<my-host-url>')
client.create_run_from_pipeline_func(
    add_pipeline, arguments={
        'a': 7.0,
        'b': 8.0
    })

However, even this tiny example lights up my IDE like a christmas tree;

I haven't done anything except pip install kfp.
My question is:

• Is this expected, and is the kfp SDK simply not typed properly, or is there something fundamental I'm missing?

I wanted to post here before I potentially submitted a ticket, in case there's just something I'm missing here.
Thanks for any and all help!

[milosjava]

We use both PyCharm and VSCode in our organization and we don’t see these hint issues in either IDE. Which IDE are you using? It’s likely something specific to your configuration.

[HackNSnack]

Thanks for reaching out!

Indeed I am using NeoVim, and when I look at the code in VSCode, I don't get any direct error messages.
However, I looked at the code in VSCode in more detail, and I think the lack of errors is just because Pylance (default LSP in VSCode) isn't as strict. The inconsistencies are still there, at least on my end.

E.g. the return types of the functions are incorrect. The decorators still don't seem to be properly typed, and you can see the type of the function changing if you add/remove parentheses to the decorator;

So my solution currently is simply to disable all type checking when using the KFP package.