On Tuesday, 6 May 2014 at 11:20:33 UTC, Paulo Pinto wrote:
On Tuesday, 6 May 2014 at 09:11:30 UTC, Chris wrote:
I recently came across this article
http://www.wired.com/2014/02/julia/. On the Julia homepage
there are some benchmarks times relative to C. I know that
bearophile has mentioned Julia several times on this forum.
Has anyone compared D's vs Julia's performance as well as
design features?
I can only comment on design features.
You can think of Julia as a dynamic language similar to Python,
with optional typing and for such a young language, a quite
good JIT compiler backed by the LLVM backend.
It is a multi-paradigm language, with an OO approach based on
multi-methods and direct support for scientific programming.
The target audience are the scientifc community that makes use
of R, Python with NumPy and so on, which are currently
disappointed with the performance of said systems. Their goal
is to keep the programming flexibility of R and Python, while
improving the performance without having to be forced to write
C code.
Excellent summary, one quibble, you omit MATLAB and Octave users
from your target audience, when they may be the most important
one. Julia reads much more like MATLAB than like R or Python, and
numerical linear algebra is one of the things Julia is being
aimed at.
It has a very rich language of types, and a macro system.
Disappointingly, whilst you can annotate function arguments and
variables with types, you can't annotate the function itself with
a return type.
TL;DR MATLAB reimagined by Lisp hackers. I like it!
Not really competing in the same space as D. Yes, I know, I'm
pigeonholing D, which is supposed to be a wide spectrum language,
etc. etc.
