You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm going to close this as we are well aware that type annotations exist and adding them across the library is a bigger project than should be tracked in a single issue. I stand by we should let numpy go through a couple of release cylces with the types before we start to adopt them to let the dust settle a bit.
It is my understanding that the MSFT stubs are auto-generated by walking the code which suggests it is not urgent that we take this on (as the tools that will benefit the most from the stubs can generate their own).
Reacted by Paul Willot
Reacted by Søren Holm, Oliver Mannion, Nate England, Shuang Wu and Jonas
It is my understanding that the MSFT stubs are auto-generated by walking the code which suggests it is not urgent that we take this on (as the tools that will benefit the most from the stubs can generate their own).
FWIW, this isn't the case; what's shipped with Pylance I wrote by hand over the course of a weekend conference (this weekend was the anniversary of me writing them, yay); they certainly aren't meant to be proprietary in any sense and will end up in https://github.com/microsoft/python-type-stubs (them not being there is an oversight).
Sorry for bumping an old thread (time flies), I'm just now spending some long-overdue time addressing issues in those stubs and was checking on the state of typing here.
Reacted by Oliver Mannion, Ole Jørgen Brønner, David Gilbertson and Shuang Wu
Activity
story645 commentedon Jul 21, 2020
The dev team may rivisit now that numpy has types. Is there any argument you'd like to add to the previous discussions #15926 & #13798 ?
tekumara commentedon Jul 21, 2020
FYI: pylance ships matplotlib stubs already, but they are proprietary (for now) and may end up in pyright see https://github.com/microsoft/pyright/issues/861
tacaswell commentedon Jul 21, 2020
I'm going to close this as we are well aware that type annotations exist and adding them across the library is a bigger project than should be tracked in a single issue. I stand by we should let numpy go through a couple of release cylces with the types before we start to adopt them to let the dust settle a bit.
It is my understanding that the MSFT stubs are auto-generated by walking the code which suggests it is not urgent that we take this on (as the tools that will benefit the most from the stubs can generate their own).
jakebailey commentedon Feb 23, 2021
FWIW, this isn't the case; what's shipped with Pylance I wrote by hand over the course of a weekend conference (this weekend was the anniversary of me writing them, yay); they certainly aren't meant to be proprietary in any sense and will end up in https://github.com/microsoft/python-type-stubs (them not being there is an oversight).
Sorry for bumping an old thread (time flies), I'm just now spending some long-overdue time addressing issues in those stubs and was checking on the state of typing here.