As you probably know, I’ve been sipping on the node.js kool-aid lately. Mostly, I’ve just been hanging out in the channel and playing with my bot.
However, my previous background is in middleweight scientific computing: Think Excel, Matlab and more recently Scipy. So, coming to an environment whose focus is on webserving is pretty different for me–cool, but different. And, to be honest, I was jonesin’ for some matrices.
I did some cursory web searches before and didn’t find anything, but after poking at the problem myself and finding it surprisingly difficult (I mean, I can do it by hand easily enough right?), I searched with more scrutiny and eventually found…
Pretty cool, right? Of course, it was aimed at being a client-side library, and not a commonjs module. But, that was easy enough to fix! So, I cloned the repo and made a few changes. This is my first experience successfully porting something, but this was a particularly easy case due to not requiring anything from the DOM.
First, if you read the Rakefile (or bumble through it like I did, if you don’t know Ruby), you’ll see that under the “build” rule, it takes the lists inside “packages,” grabs all the files in src/ with that name, and concats/minifies them into a file. So, I just added a second rule:
#snipped from the Rakefile
PACKAGES = {
'sylvester' => %w(vector matrix line line.segment plane polygon polygon.vertex linkedlist),
'sylvester-common' => %w(vector matrix line line.segment plane polygon polygon.vertex linkedlist common)
}and added a javascript file that would add the commonjs exports:
//common.js
module.exports = {$V: $V,
$M: $M,
$L: $L,
$P: $P};Then, I mashed “rake” and had an extra file, “sylvester-common.js,” which could be imported in Node!:
node> var m = require('./sylvester-common');
node> var a= m.$M([[1,2],[3,4]]);
node> var x=m.$V([1,1]);
node> a.multiply(x)
[3, 7]Pretty sweet, right? I mean, that’s a trivial example, but it shows that it works, and if you check out the docs you can see that the basics are covered pretty well.
The only downside (and really it’s a two-sided coin) is that this is a pure javascript library. On the plus side, that means the un-common version will work in the browser great! (Anybody else seeing some mapreduce potential using dnode here?) On the minus side, there’s no way it could go toe-to-toe, performance-wise, with any “real man’s” library, making use of optimized c/c++/fortran codes such as lapack, BLAS, atlas, etc., like numpy does.
In other words: ymmv.
If you want to try it out, my repo’s here. For convenience’s sake, I git added the built libs. Enjoy!