What would you want the most in Python 4?

Now that python 3 has hit 3.8 we can expect that python 4 will come out pretty soon.

Please put in the comment the most important single feature that you wish python 4 will have.

To start, for me it would be a stricter typing: typing is currently just a hint and not enforced in runtime. In a big project it helps a lot to have typing for everything (variables, functions).

Would be perfect if python can have a static typing mode that enforces this.

Comments

[Nguyen Kim Son]

I think the function here is rather canConvertToInt (or can_convert_to_int according to PEP8 😅).
The int(val) works for both val=1 and val="1". If you want simply to test whether a variable is int, type(val)==int should be enough.

[Eric Ahnell]

Python 3.8 gets a heck of a lot closer by putting the relevant typing module in the standard library, so you can leverage it much more easily. That being said... the biggest thing Python needs right now is architectural improvements in multiprocessing. So many things people do with Python benefit from multiple cores, and Python can't yet take full advantage of them.

[Nguyen Kim Son]

In order to take advantage of multi core, it seems that Python developers often use native binding libraries like Cython and hardly take use of the multiprocessing package. For IO bound applications, asyncio is usually the solution.

[Pykash]

fastest as java.

[Nguyen Kim Son]

To speed up Python, Cython can be a good option as we can mix C and Python code. We used to call 3D C++ functions from Python with Cython and the performance is on par with pure C++.

[Younes]

i++ !

[Evan Roggenkamp]

Hey you might be interested in mypy

[Casual Coder]

Get rid of GIL

[Jason Held]

Have you had experiences where the GIL was the limiting factor on the runtime performance?

There are PEPs open now to help with this (including an overhauled subinterpreters implementation).

[Casual Coder]

No, never reached that stage honestly

[Fulton Browne]

I think python 4 along with any other new version of a programming language should have multiple targets like kotlin and haxe have.

[Nguyen Kim Son]

They are indeed similar. I think Cython users will tend to write custom C code whereas JNA is more about using existing code without modifying it but I might be wrong.