Yay, I think I fixed all the issues! So feel free to use the [tex] tag to make all your math look awesome! And as a sample:
##
\begin{equation}
\label{xx}
\begin{split}a_{1}& =b+c_{2}-d\\&
\quad +e-f_{5}\\&
 =g+h\\&
 =
\cfrac{z}{\sqrt{2}+\cfrac{1}{\sqrt{2}+\cfrac{1}{\sqrt{2 + \cfrac{x}{y}}+\dotsb}}}\end{split}
\end
{equation}
##
Which was made from
\begin{equation}
\label{xx}
\begin{split}a_{1}& =b+c_{2}-d\\&
\quad +e-f_{5}\\&
 =g+h\\&
 =
\cfrac{z}{\sqrt{2}+\cfrac{1}{\sqrt{2}+\cfrac{1}{\sqrt{2 + \cfrac{x}{y}}+\dotsb}}}\end{split}
\end
{equation}

Did you or albert ever move it, or is the IRCD still in my home dir? Cause I just left it there when I set it up oh so long ago and last time I checked it was still there Probably not the best place to store that I feel though, especially since I don't exactly manage it anymore.

haha, isn't it? And 9223372036854775807 (Long.MAX_SIZE) is the largest possible seed, though theoretically it should let you type in any size of number, it just might crash when you click change

Scipi: Nice one! What seed is that?
Juju: Huh, yeah, that looks like an internal jre error. 

So a couple days ago, I randomly thought of how awesome procedurally generated music would be, and when I looked, there is a (in my opinion) surprisingly small amount of these kind of things. So I decided to write one for fun. Right now, it is very very random sounding and it doesn't really flow at all, but it can still be kind of cool sometimes still, especially when you keep in mind it's all randomly generated I just started this, so it is very early, and will probably change a lot in how it sounds, but this is a very early showcase of how awesome this can be. Right now it loads up with the seed 666 by default, but you can change that. Everything about the song is randomly determined based on the seed. The program is written in java, and attached below. So try it out, and maybe post some of your favorite seeds and stuff!
Note: Large seeds work better!

Now, for those interested, here is a quick overview of he current process:
1. First, it chooses a random tempo (50-170bpm) and a random root note.
2. Based on that root note, I have a formula which transforms the set of integer numbers into the notes on a major scale based on the root note
3. For each measure, it first chooses a root note for the chord for the measure (based on the aforementioned formula), and plays the triad based on that note and in the scale for the root note. Note that I took out it choosing the seventh (a diminished chord) for now because when randomly stuck in there, most of the time it just sounds odd.
4. There are 32 ticks, or steps, in a measure, and each tick is equivalent to a 32nd note at the chosen tempo. So it then figures out how to fill up those 32 ticks with random length note from the chosen triad, varied randomly by octave, choosing lengths between half notes and thirty-second notes.
5. There is also a simple static drum beat, just to keep the sometimes very weird off-beat things that can happen from getting too disorienting
And that's about it for the current version! I hope to start adding things that make the notes and measures more dependent on each other to really make everything start going together better, and also add some generic formulas for some additional melodic parts to the song. There's also quite a few bugs I know of, but for now, enjoy!

NOTE: CHECK LATER POSTS FOR THE NEWEST VERSION.

Java. Here is the com.up.ti package. This is the source, so if you can't figure out how to directly use, you can maybe just extract the read/write routines. Or if you need more help using this just ask.  Note: There is the PictureFile class in there, and also a Real class, they arent actually needed for the program stuff, but theres some references that would break if i took them out, so i left them (actually, there are quite a few excess classes that are there for multiple file-type stuff. if they really bug you, i can give you a version without them)

To read a file:
ProgramFile pf = new ProgramFile();
pf.openFile("Cool File.8xp");
pf.Tokens //String-ization of all the tokens
pf.Name //Program name
//To get a short[] of all the bytes of the program:
pf.ts.stringToData(pf.Tokens);
//To load a custom token set (see the file syntax in the main tfe thread)
short[] temp = pf.ts.stringToData(pf.Tokens);
pf.ts.changeTokens("Custom Tokens.tokens");
pf.Tokens = pf.ts.dataToString(temp);
//Or easier
pf = new ProgramFile();
pf.ts.changeTokens("Custom Tokens.tokens");
pf.openFile("Cool File.8xp");

So have fun!

Also: Credit would be appreciated

First of all: There aren't really any "data types", everything is all about byte arrays and how you use them. And now that I've said that, let me tell you what the data types are
There are 8-bit or 1-byte numbers which have a signed range of -128 to 127 (equivalent to a java byte) or an unsigned range of 0-255. Note that unsigned and signed aren't really any different, it's all about what commands you use with them.
Then there are 16-bit or 2-byte numbers which have a signed range of -32,768 to 32,767 (equivalent to a java short) or an unsigned range of 0-65,535. There are also 8.8 fixed point numbers, which are also 16-bit, but have a signed range of -128 to 127 and also have a fractional part with a resolution of 8-bits, for representation of decimal numbers. These are not the huge range floating point numbers you are used to, but they can still be very useful at times.
Next, about the guide, have you read the pdf included in the axe zip?