3 ms·
Were you a Python-convert? Could you share you transition journey like how much time or difficulty you faced.
by amrrs 7y ago
Were you a Python-convert? Could you share you transition journey like how much time or difficulty you faced.
- speps 7y agoIn my own recent experience, Julia is awesome but documentation could be sometimes clearer (eg. missing examples with each method, missing cross references). Also, when searching for details on some features, it's hard to figure which version someone is talking about, it seems some versions (0.6 for example) were there for ages and most forums refer to it but some stuff is now deprecated or replaced. Otherwise, I would probably use it for everything now given the performance and the very good FFI story.
- ViralBShah 7y agoPlease file issues when you find something is less than perfect. That's the first step towards making it better!
- adamnemecek 7y agoThere were some problems during the transition between 0.6 and 1.0 but that was kinda expected. After 1.0 it's been pretty breezy. Good IDE support (juno). Julia is fast. Numpy is actually c, julia is just julia. You have an amazing interop that just works. With python, matlab, c,c++,etc etc. https://github.com/JuliaPy/PyCall.jl https://github.com/JuliaPy/PyCall.jl Package manager just works. The language is a lot more powerful. Diffeq is around 1000 lines of code which is nuts.
- ChrisRackauckas 7y agoDiffEq is a lot more than a 1000 lines of code. By now it's about 75 git repo and some of those repos are themselves >10,000 lines of code. And there is a single file in the test suites that is 10,000 lines defining mathematical constants for 200 Runge-Kutta tableaus. So it's way way more, and actually quite large by now. However, we still get a lot of feature bang for our lines of code buck.
- adamnemecek 7y agoLol. I think I looked at it a while back.