Hacker News Reader

That's not the same as Google's support for Python, Go and Kotlin or Microsoft's support for Typescript and maybe soon Rust. I would have thought the case for supporting Julia was very strong considering Python's performance and concurrency problems.

View on HN

Python's performance is less of an issue when you use libraries like Numpy that are basically running optimized C code.

Depending on what you're doing, Python is usually great at performance where the time it saves the user writing code is more valuable than the performance itself (a lot of exploratory scientific research).

Numpy is great if you can vectorize your code in a straightforward manner. But when you start doing anything non-standard it becomes difficult.

I can tell you that programmer productivity can plummet drastically when you need to get everything into vectorized form.

Also, custom classes and types do not vectorize well at all.

Agreed.

Reply on news.ycombinator.com