Become a fan of Slashdot on Facebook


Forgot your password?
Last Chance - Get 15% off sitewide on Slashdot Deals with coupon code "BLACKFRIDAY" (some exclusions apply)". ×

Comment Re:So . . . (Score 1) 106

Because Julia is compiled into the same bit code that C is compiled into when compiled with the Clang (Apple) compiler. That means no translation necessary for calling to and from C code (and soon C++). When Julia gets slower it is because: (1) the garbage collector is immature, and (2) you are using facilities that simply aren't there in C.

Comment Re:So . . . (Score 3, Informative) 106

There will be a big "ease of use" benefit to using Julia though. Julia requires to bindings to run C code (and soon C++ code, including all those fancy templates). That's because Julia doesn't suffer the two-language problem that every other language does. Julia types are LLVM types, and C types are LLVM types. Same with function calls. It's so simple and convenient.

Comment Re:So . . . (Score 2) 106

No they are not. Python has a two-language problem, for a starters. Also, Python is a pain in the neck for scientific computing, which is my day job, btw. Also Julia is soon to work seamlessly with both C and C++ (like, include header, load library, and execute code without any bindings or nonsense). Julia has the potential to be properly threaded, but is really aimed at being seamlessly multi-process. That mean seamlessly moving data between processes -- even on different computers and networks. This is seriously useful for scientific computing. Julia is much more like Matlab, and Matlab is much, much, much better for scientific computing than python ever can be. Also, Julia should work faster than python once the garbage collector has been updated.

Comment Re:Do we need another language? (Score 2) 106

Julia is revolutionary, and I'm looking forward to it blowing away Matlab. All that's really holding it back is a good garbage collector, and mind-share. The language is truly superior, integrates seamlessly with C (like it is the same language -- there is no two-language problem), and soon C++. Could seriously replace C++ for /most/ things if it were multi-threaded. (There is some work on this, but Julia is designed to be multiprocess -- you can seamlessly shift objects between processes, even if they are on different machines. This is better then threads if you're doing scientific computing.) So Julia even has the potential to be better than D. And it runs as fast C (compiled by clang), and comes with the convenience of a REPL, and a truly beautifully designed.

Julia looks a lot like Matlab, and some Matlab code just runs as is, or requires only minor tweaks. Mathworks has long dominated scientific computing with their awkward and expensive -- but otherwise so convenient -- product. Mathworks is screwed.

Java/C#/python will never be what Julia is.

Comment Re:In other news.... (Score 5, Interesting) 500

The minimum wage in Australia is much higher... about $US15 per hour depending on which "accord" (industry) you are in. Having lived and worked in the US, Canada, and Australia, I can attest that minimum wage earners *work* about 4x harder than they do here. You see, hundreds of people line up for jobs that they then try to *keep*, since you earn about $US30k per year on them. (That is about the median US salary.) And the unemployment rate is comparable or lower, and the debt is less, in part because there is less need for social services. (Australia is one of the lower tax OECD countries.)

This situation arose by a law passed in the early 80s that made it illegal for unions to campaign for pay raises without showing an increase in productivity. Businesses, in turn, had to pass on some of the increased earning from productivity gains. All of a sudden, we have unions and businesses on the same page, with unions responsible for their own worker productivity, and the amount of hours-per-year lost from industrial action was an order of magnitude lower than any other OECD country.

Neoliberal economics gets a lot right, but there is a flaw in its theory surrounding labour law. People are not replaceable units, and workers are "sticky", in that they have families and other commitments. This is not true for some industries (like some types of internet work), but it is mostly true. This sets up a very big power differential between businesses and workers, and a type of "prisoners dilemma" where individual businesses act in a way that is helpful to themselves but detrimental to the aggregate.

Building translators is good clean fun. -- T. Cheatham