After careful statistical analysis I concluded that there is a 9.56% chance Zed is trolling experienced programmers. Disregarding the trolling about "Turing Co

After careful statistical analysis I concluded that there is a 9.56% chance Zed is trolling experienced programmers. Disregarding the trolling about "Turing Completeness", there are some points he raises that are interesting. On some (most thing) items there, the train has left the station. But some are good to listen to: Better error messages. Rust has been making progress there so maybe a good place for inspiration. Error codes and links to additional info online has been done (maybe overdone) but I think there is room for more variable names, more info about the context ("looks like you're trying to do X, maybe Y might be better here..." kinda thing). On the lack of typing and such, MyPy might be a good project to focus on. Standard library updates would be good, but they are tricky. Wouldn't be nice to have requests in the main library? Yes, but changes are still made to it, it might not be updated for a long time then. As they say modules go to the standard library to die. Things I'd like to see: better performance (close working with Pyston or PyPy) and better packaging. On packaging there was talk of Pipfile just today. That was a good start. I've been using Python for many years, and I still don't know off the top of my head the basic pip flags and options, or how to start the venv, what goes into setup.py vs goes in requirements.txt vs goes in setup.cfg. The story got better in recent few years, but there is more room for improvement there I think. I often end up with a build.sh script or run.sh which just does the building and venv-ing and such. Can that be done by default? For usability sake? Even if it is not super-consistent and doesn't work for 1% of use cases... Probably.