I am not enjoying the way #mypy can suddenly find new type errors in code that hasn't changed at all. #python

(but only in CI! so if i add a comment to ignore the ‘error’, local mypy — the same version — complains that the comment is unnecessary.)

Is there any *good* type-checker for python? But I really mean a good one... I had bad experiences with mypy being not smart enough and even pyre-check can't handle `Optional` types correctly (similar issue with mypy).

Should I just give up on trying to have a better experience with python and just accept the fact that it is an inherently bad language to type-check?

#python #mypy #pyre #typechecker #typehints

Yesterday mypy forced me to learn about covariance, contravariance (and by extension invariance) in the context of programming language typing which isn't exactly what I signed up for, but it's hard to complain about learning new things so ¯\_(ツ)_/¯

https://github.com/zcutlip/pyonepassword/blob/07fb23dc3b381a711a90e93b88a00cdd6e107a72/pyonepassword/_op_cli_argv.py#L368C13-L368C13

#python
#mypy
#ComputerScience
#programming

No więc dzisiaj aktualizuję #mypy do 1.5.1, i widzę, że nie ma żadnych zmian w kodzie, tylko zmiana numeru wersji. Mały CDF (co do faki?).

Teraz ztrawhcse gada mi na IRC-u, że przecież zmiany są:

https://github.com/python/mypy/compare/v1.5.0...v1.5.1

ja na to: no tak, czyżby nadpisali tag? Więc wymuszam ponowne pobranie, ale suma kontrolna się zgadza. Coraz większe CDF. Pobieram przez przeglądarkę, dostaję ten sam plik.

Wtedy ściągam 1.5.0… i mamy facepalm miesiąca. Nadpisali 1.5.0 na wcześniejszą wersję kodu już po tym, jak zaktualizowaliśmy paczkę w Gentoo.

Innymi słowy, użytkownicy Gentoo właściwie używali 1.5.1 już od tygodnia. Nie musicie nam dziękować.

#Python

@folkerschamel @ado

Asking for specific #examples type checking finds is ... unproductive. You write 2 modules of code, run #mypy, and get a list of the 11 places you need to fix - before you've even written unit tests that may or may not have caught the same problem.

If I tried to keep track of them all, I wouldn't have time to write code.

[...]

#TypeHints #TypeChecking #tests #lint

@kellogh In #python, #mypy already reports not awaited functions that you don't assign anywhere. And if you do assign, you'll get a type error when you try to use it as a value instead of as a coroutine.

@EMR @astraluma I know a few tools!

https://github.com/typeddjango/awesome-python-typing#testing

The difference in these tools is where you want to put your type tests: next to regular tests, in yaml, or in a separate Python file(s). The first approach is bad if you don't want to exclude all your tests from type checker. The second one is fine but you lose autocomplete and linters. I'm the author of the third way:

https://github.com/orsinium-labs/mypy-test

Also, basic tests are possible with just #mypy:

https://pythonetc.orsinium.dev/posts/assert-type

#python

@ambv so, what do you think about this problem?

It also behaves the same way in mypy. But here bug is not closed, and i'm hoping it can be fixed.

https://github.com/python/mypy/issues/11583

#Python #PythonTyping #Mypy #Pyright

#Python #PythonTyping #Pyright #Pylance #Mypy

Is there a way to create type stub only for part of the library?

For example: I want to include types for a single object, but don't want to maintain full copy of the library.

https://github.com/carltongibson/django-filter/pull/1585/files

Is it me or #Mypy just got a whole lot faster? It feels like it runs as fast as #Ruff now! 😄

I really enjoy #mypy updates: transparent, regression-free, and each time it seems #mypy understands more things about my code, so I can remove some type-ignore comments. Kudos to #mypy's maintainers and contributors!

@python_discussions That's not true. You need to specify `__all__` in `__init__.py` (and other facades) so that #mypy, #flake8, and #ruff know that these packages are imported to be exported. Then flake8 and ruff won't report unused import and mypy will allow reexports from the module.

#python

@glowrocks That's great! I'm glad to hear it empowers you to achieve your goals.

If you need guidance on more classic and reliable tools, just ask. In #python, I recommend to have a look at #ruff and #mypy, if you haven't already. These are good for finding simple errors without running the code. And Deal and Hypothesis should help with generating tests.

@benc mypy will report it by default, unless the name you are trying to import is aliased in the proxy module using `as` or included in `__all__`:

https://mypy.readthedocs.io/en/stable/config_file.html#confval-implicit_reexport

Edit: reading the docs again, I see that reexport is allowed by default, you have to explicitly forbid it in the mypy config.

#python #mypy

Just saw that there is now an official VSCode extension from Microsoft for using mypy.

https://marketplace.visualstudio.com/items?itemName=ms-python.mypy-type-checker

Very cool!

#python #mypy #vscode

Having recently tested #SonarQube I find my #black #isort #flake8 #mypy toolchain for #python does 98% of the code quality job, the rest being covered by code review. What do you think and what do you use ?

Anyone out there having to support more than one version of MyPy|TypeShed|mypy-extensions|python with the same codebase?

Even with having different .mypy.ini's sometimes the hints in the code are incompatable and you end up playing whack a mole.

For example, is it r: Redis or r: Redis[bytes] or Redis[str]?

#mypy #python

#TIL executing #mypy as a daemon with an sqlite cache makes it go really fast. Squashed a lot of type errors with that workflow.

https://mypy.readthedocs.io/en/stable/mypy_daemon.html

https://mypy.readthedocs.io/en/stable/additional_features.html#caching-with-mypy-daemon

dmypy run -- --sqlite-cache --use-fine-grained-cache <path>
dmypy check <path>

#python #typing

📢 Announcing v3.0.0 of our #github reusable #workflows

https://github.com/coatl-dev/workflows/tree/v3.0.0

#github #black #flake8 #pylint #mypy #python2 #python #nox #tox #precommit

Is there a common way to typehint slice? Especially for cases, when you want specific things like label-based slicing (think Pandas `.loc` for a string index). Pandas does not have any typehints for keys afaik. #pandas #python #datascience #mypy