I was just saying that getting really good performance out of LLVM with Julia isn't always obvious. This isn't just a Julia thing. People starting with Python +

I was just saying that getting really good performance out of LLVM with Julia isn't always obvious. This isn't just a Julia thing. People starting with Python + Numpy have issues with speed when they don't take advantage of the vector primitives and drop into regular Python loops for part of the code. First time Haskell users are frequently confused when it is significantly slower to parse text than Perl as one of the default string libraries is really slow and not recommended even though it is what a new user would think should be the first choice. Overall I'm excited for Julia, but have a lot to learn. As far as Octave/Scilab type applications go, I find them useful for small systems where I need to make sure the math of what I'm doing is correct before writing it in something like Julia or Python + Numpy. Excel is also surprisingly good for the really simple stuff (Ex: inverting a small matrix and doing some matrix multiplications).