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
Given the growth of the source code, tracking inputs/returns of functions/classes is not smooth, mainly when working on a very connected code. I think it will be super helpful to start adding type hints while developing something new or editing something old.
I'm at +/-0 on this - I intellectually understand that they can be valuable, and this case study is pretty convincing, but I think they also raise the bar for code contribution even more than tests do. And from the case study above:
This effort took many months for our team to complete so make sure to allocate enough time.
...I would like to point out that we are not exactly replete with developer time, so, you know, let's be careful, eh?
So a few miscellaneous thoughts and feelings:
I absolutely am not in favor of a large-scale effort here. We should add them gradually, if at all, and make sure we build out the practice of adding them.
there are "internal API" modules and classes that I suspect would benefit the most. Think search.py, compare.py, manifest.py, picklist.py, signature.py, minhash.py, and so on - but not commands.py, or the code under cli/.
Given the growth of the source code, tracking inputs/returns of functions/classes is not smooth, mainly when working on a very connected code. I think it will be super helpful to start adding type hints while developing something new or editing something old.
References:
The text was updated successfully, but these errors were encountered: