Hi,

On Wed, Feb 19, 2014 at 2:13 PM, bp2012 <[email protected]> wrote:
> I'm not a scientific programmer. In fact, I haven't been a professionally
> paid developer/dev mgmt for 8 years. However, being able to quickly code up
> a solution to a problem, to perform my own research on our data, and
> generate my own reports has proven to be highly valuable to my career on the
> business side.
>
> With that said, I have been on the hunt for a language that is easy to use,
> allows me to concisely expresse what I'm looking to accomplish, and have
> reasonable performance. I've used verbose languages like java and c#,
> scripting languages like perl and python, next generation clr based
> languages like Nemerle and Cobra, functional languages like Lisp and
> Clojure, and even self contained environments like Pharo smalltalk. There
> are great features in those languages, but there was always something that
> after a couple projects that jumped out as a big turn off or an obstacle
> that couldn't reasonably be overcome. In each case, I jumped ship and moved
> on to find the next promising candidate.
>
> I've been using julia for a number of months and have written several
> projects. I haven't found a single obstacle that couldn't be overcome. With
> ccall and pycall I can access a ton of libraries and the overall julia
> syntax is so concise and expressive that writing software is enjoyable.
>
> I'm convinced that julia has a huge future in non-technical / general
> purpose software development. But since not a lot of effort is expended
> (yet) on publicizing to the masses, only the people actively searching for
> an alternative will find it in the short term.
>
> To be honest I think that is a good thing. From my perspective, the
> development efforts are rightly focused on the important things. For the
> language to appeal to the masses, the installation process and environment
> configuration would need to be idiot proof and a ton of introductory
> documentation is needed. Currently, if you run into a problem you might have
> to actually research it yourself.  You will have to read source code that
> you didn't write and maybe even have to make your own temporary fix. This is
> a non-issue for motivated, bright technical people. However based on my
> experience, it's a non-starter for an average business applications
> developer.

+1 to everything you wrote above. I'm a scientific programmer who has
been using Python thusfar, and although Python has many mature
libraries (and some newer ones), I find myself using Julia for
experimental projects. While its interesting to poke around the source
code and have moments of epiphany, I grok this may not be suitable for
teams that want production grade code (now).

The only installation problem I had was hardware (laptop heated to 98
deg Cel. and shutdown) issues with BLAS compilations on an older
laptop, so after that single compilation, I stick to @staticfloat's
nightly builds for Ubuntu. That said, despite the compilation issues,
I still prefer "git" over packaged versions - you will know what I
mean if you've ever struggled with installing python scientific
packages inside a virtualenv. That was many moons ago but the
dependency hell was truly hellish.



> As the environment and packages mature, these issues will go away and over
> the next couple years, my prediction is that julia will seen a massive
> increase in adoption outside of the technical computing world.
>
> I want to do my part and blog about the general capabilities of julia and
> some of the excellent packages such as DataFrames, ODBC/SQLite, Match, and
> PyCall, that make developing quick and painless. It's just a matter of
> finding the time ;)

I try to collect interesting cookbooks here:
https://github.com/svaksha/julya/blob/master/Cookbook.md#blogs
When you find the time to blog, feel free to submit a link (as a PR)
or file a bug and I'll add it.

- svaksha

Reply via email to