I used Julia for both a machine learning course and a neural networks course this past semester, and I really enjoyed the experience of developing in the language. I found it really a lot faster than python/numpy and a lot more sane than Matlab. Some key syntactical structures are also very purposefully kept similar to Matlab so it was very easy to translate code between the two, which is incredibly useful in scientific computing. I definitely recommend people check it out; however, I will warn any potential users that the run times of Julia programs right now (as of 0.3.x) are incredibly slow if you need to include any third party libraries. I understand they're working on this for a future release (maybe 0.4), but even including a simple plotting package in your code currently causes a 20-40 second overhead before your code actually starts running because the packages aren't precompiled.<p>Oh, and one more tip for OSX users wanting to try out Julia. You might want to just use the app bundle provided on the Julia homepage rather than compiling from the homebrew tap. I spent close to an hour waiting for all the dependencies to compile before giving up when I realized I don't want to be doing this every time it's updated.
For a while now I've been interested in learning more about Julia, though not exactly sure yet what use-case I have for it. I guess I wouldn't know until I do study the language enough to figure it out.<p>This tutorial looks promising, building knowledge carefully but not too much "hand holding", like it's for programmers experienced in other languages. Got part of the way though it so far, obviously not a "one-shot" kind of exercise.<p>My only complaint is the same one I make with many other tutorials of this kind. I find it very annoying when some feature or function is used in an example, but the function comes "out of the blue", not defined or explained, as though the reader is supposed to magically divine its meaning or definition.<p>In this case, everything goes along swimmingly until encountering the function/method "repr(a)" in "Arrays" examples. I dutifully spend several minutes tracking it down, and what do you know, basically the same as (->string ...) in Scheme (i.e., object->string representation).<p>I'm not giving up on Julia. I'll carry on from that point when I return to the site. But really, IMO <i>every</i> newly introduced term/function/feature should be defined so that the example can actually be completely understood.<p>Otherwise there's a substantial risk of losing readers who will not so likely be motivated to keep trying to figure it out.<p></end of minor rant mode>
I don't do any scientific computing, as such I have zero use cases for Julia.<p>However as language geek, I do follow its development.<p>It is quite interesting to see another dynamic language in the same footsteps of Lisp dialects and Dylan, where performance, ability to compile to native code and simple FFI are considered important features to be part of the main implementation.<p>Really looking for Julia becomes widely used.