Julia is a fast, dynamic, composable, general purpose programming language well suited to statistical data analysis, machine learning, automatic differentiation, other numerical modeling, as well as more general purpose programs. It feels like a scripting language, but runs compiled -- using Just-In-Time (JIT) compiler technology to compile code on the fly. It was designed from the ground up to support multi-threading, multiprocessing, and distributed multiprocessing. Julia solves the "two language" problem -- you don't need one language for ease-of-use prototyping, and a second language for speed -- as Julia is already easy to use and fast. You can write your high performance code in Julia itself.
This proposal would be to support Julia similar to how Python is currently supported.
Data types are highly compatible between the two languages
| JMP Data Type | Julia Data Type |
| Boolean | Boolean |
| Data Table | DataFrames.jl |
| Associative Array | Dictionary |
| Numeric | Float64 |
| Matrix | Float64 Matrix |
| List | tuple |
| String | Unicode string |
One advantage of Julia over Python is that there is no Numpy -- that functionality is just built-in.
In addition, instead of telling users what packages they need to install, JMP could use PackageCompiler.jl to create a stand-alone binary of Julia with all of the packages that JMP requires and distribute that with JMP. Julia integration would work right out of the box (or "after the install program finishes"). Advanced Julia users would have the option of replacing it, but for everyone else, it just works.
Julia is an exciting, rapidly growing language in statistical, machine learning, and scientific computing. See The unreasonable effectiveness of the Julia programming language, by Lee Phillips (10/9/2020).
Adding support for Julia would also users to drop some state-of-the-art tools and coding into their JSL scripts.