open-source in 2012. • Originally seen as replacement to Matlab ®. • It uses LLVM to generate “genuine” machine code. • It has run time macros for creating boilerplate code. • It has an extensible, flexible type system. • Connectivity to shared libraries has zero-overhead. • Also it can call Python, R, Java etc. • Code is on Github (& documentation on Readthedocs). • I/O parallelism is novel and offers distinct advantages over Hadoop, Storm & Spark.
yet reached v1.0 … • … and is not likely to get there until well into 2017. • v0.4.3 is stable and v0.5 in development. • There are currently over 850(+) registered modules. • These are grouped in “communities” such as JuliaStats, JuliaOpt, JuliaDB, JuliaWeb … • Version segmentation (v0.3) and module caching (v0.4) have now been implemented. • Jupyter IDE is familiar to Python developers. • Visualisation (graphics) are not built-in. 3
formed mid-2015. • Julia is a NumFocus project. • v0.5 due 1st April, v0.6 in October. • Arrays, tuples, data frames and d-arrays are being rejigged (Arraymageddon). • New data types are being introduced. • Provision for custom cluster managers and shared memory access. • Better package management using libgit2. • A Julia debugger (Gallium.jl aka GDB). • ARM versions: Raspberry Pi2, Crouton, Scaleway. 4
of Julia as the new kid on the block. • It remains one of the most “democratic" open source programming projects. • Julia Computing has attracted considerable funding and is undertaking consultancy work and still includes all the three original developers (+ others). • Strategy seems to be still to increase ‘core’ functionality upto v1.0 rather than just depending on package support. • For academics and SME’s it offers some interesting alternatives, for Enterprise work it is more debatable. • Python and Julia fit well together. 5