Julia is very interesting! I started playing with it again using the Juno plugin for Atom and it’s been fun. Prototyping in the REPL seems like a must to be productive.
I wrote some code for an early version (0.5?) some years ago and man the compile times were killing me. Just importing a plotting package took around 30 seconds. Luckily things seem to have been steadly improving since then.
That assembly view with the whole… inlining stack (?) is impressive. I want that in all compilers.
Julia has grown into my all-time favorite general-purpose programming language. It beautifully combines the accessibility of Python, the power of Lisp, and the performance of C, all while introducing its own set of mind-blowing magic tricks. I’m surprised that it hasn’t already taken the world by storm. If you do a lot of deep prototyping or scientific computing, you absolutely must give Julia a try.
I recall @dl gave it a try in December 2014 and had some stuff to say about it. It might be interesting to follow up on some of that, nearly 5 years later.
Julia is a great language in many regards, and very promising, but if it’s going to unseat the incumbent data science languages (I mean Python and R) it’s going to need a lot more libraries, and even then it’s a long uphill struggle against network effects from entrenched codebases and mindshare. Consider how much (most?) software is still written in C++ despite the long existence of “obviously better” alternatives. We can hope for a “sea change”, but don’t bet on a “storm”.