Python Domination Threatened by Julia?
thenextweb.com
thenextweb.com
First, no, Julia is not threatening Python any time soon. _If_ Julia ever overtakes Python, it will take a decade, at least. The inertia is just too large, and it's exacerbated by the people who use Python: Beginners who just want a big community and plentiful learning resources, and people who need large, existing libraries. These libraries and tutorials take a long time to make, and big companies haven't even _started_ investing in Julia. They probably should, and I hope they do, but in any case, it's way too early to imply Julia threatens the crown.
Second, the whole "Zen of Python versus the Greed of Julia" is misleading. All programming languages make tradeoffs in their design, Julia included. Julia has managed to settle on a design with remarkably many advantages, but there are still many drawbacks. Just go to any Hacker News thread on Julia and watch people complain about all the things Julia doesn't do well.
Third, it annoys me that some people keep saying Julia is faster than C. No, it isn't. It's generally slower, but in the same ballpark. It's fast enough that you never have to go to another language for speed, but it's _not_ faster than C.
Then there's this quote: "But in contrast to Python, you can introduce static types if you like — in the way they are present in C or Fortran, for example". This bugs me, because Julia is still dynamic, even when it has type annotations. It's just throw an error if the type annotations are incorrect. To me at least, that's not static in the same way as static languages.
Look, Julia is great. About one year ago I switched completely from Python to Julia at my job, and it's been a nice experience. But there is little use for these low-effort posts that don't really persuade anyone.
But that's definitely good enough for an awful lot of numeric work.
Personally, I kinda like Julia because it's a little closer to R, and I hate hate hate doing analytical work in Python (ML/programming in the larger is fine).
I do agree that this is a terrible blogpost though, but Julia does have some advantages.
I wanted to start evangelising it when it was 1.0, but that was not a great launch (IMO, obviously). They did it a little too quickly, before all the library authors had updated.
I never got paid for writing Julia, nobody I know that used Python ever switched to Julia.
Only some niche research groups uses Julia in some universities.
Nowhere that I personally know off uses Julia in industry.
And I see no reason to use it either. I use PyTorch, fastai, and sklearn for work and play. Analysts uses R (and Excel at a previous work). Julia has no usable library to be used for practical, dependable tasks. Last time I checked, the pre-compile times for even the simplest functions were painstakingly high.
I have started to dabble with Jax/Flax, and the PyTorch XLA library. Amazong things happening in the TPU space. And it all uses Python.
A research group (in Astronomy) switched to Julia, but they used FORTRAN and MATLAB.
I don't see Python's domination threatened by Julia in any way.
Python, however faces competition from Go in network programming, and Rust for CLI programming.