RREF stands for Row Reduced Echelon Form, and Here is a good tutorial on how to do it, and gives a good use for it.

edit: and if you're very familiar with matrix operations and terminology, here's a more advanced article.

It has to have one extra page to support 2.53 MP
From what I've heard however, it supports all 8. Curse you TI! Why can't you support it on real hardware!

that would be much too convenient

{B*2+{L3}}{^r}{dot}{B*2+{L6}→C}{^r}→{C}{^r}
see where it says {B*2+{L6}->C}? well, L6 is the area of RAM that is the buffer. the buffer has to have an address in memory. so when you store B*2+L6 to C, that makes C "point" to the address of the buffer. then when it says ->{C}{^r}, it stores to the two bytes of where C points, which is the buffer.

wednesday works best for me.

you're somewhat right. you can treat A-Z and theta as normal vars as you would in BASIC (just don't go over 65535, negatives are a little trickier to figure out, and don't try to store a floating point). if you do 5->A, A will be a pointer to the RAM address 5. it's not really necessary to understand this. but you should know that doing Expression->{A} is generally bad, because you're overwriting RAM that the OS may need to function.
curly braces are your friend, if you're using L1-L6 and static pointers (GDB, Pic, Str). quick code example. say you have a list of 5 numbers. 2,12,38,25 and 3. how might you create this list?
.HEADER
[020C261903]->GDB1    .Stores the data as hexadecimal
Data(2,12,38,25,3)->GDB2    .Stores the same data in GDB2.
For(A,0,4
Disp {GDB1+A}>Dec,i       .imaginary i. GDB1 can be replaced with GDB2 for the same results.
End

alright. for starters, what are GDB1 and GDB2? they are static pointers. they basically mark some data. each piece of data is only one byte (two bytes is more advanced). curly brackets return the byte that the pointer points to. this is another way of saying that a curly bracket takes the marker (GDB1) and returns whatever is inside of it (2). since data is stored linearly in RAM, we can add one to GDB1 to get the next byte. {GDB1} will return what's at GDB1 (which is 2) and {GDB1+1} will return the next byte (12).

hopefully that helps. it takes awhile to get used to.

and yes, when you're using external variables, strings and appvars are the easiest to deal with.

3) yes. FF is one byte. F is a half-byte (also known as a "nibble"). a bit is either a 0 or a 1. a byte is 8 bits. meaning, a byte is 8 1's or 0's. since there are two possible values a bit can hold (a 0 or a 1), and a byte has 8 bits, a byte can hold 2^8=256 different values. 0-255.

4) in the Commands document, the lengths of L1-L6 are defined. using L1 is the safest to use, so i'd say that L1 is probably the best to use.

5) i kinda explained this above, but to reiterate. A-Theta are pointers, but the RAM they point to can be changed. Str, Pic and GDB are static pointers. that means you tell them what data to point to at the beginning of the program, and they will point there throughout the program. using curly braces around Str, Pic and GDB is definitely ok.

6) curly braces are used to get a value from a pointer. it's very useful for lists of numbers, such as map data. i'm not sure what other things pointers can be used for, a computer/ASM programmer probably uses them more creatively.

just wanted to add in that with external var support, you should generally stick with storing to Strings and appvars. lists, matrices and real variables have complicated formats so they can hold large numbers and decimals.

as for pointers, maybe this will help. imagine RAM as like a train. and each segment of the train is numbered: 0, 1, 2, 3, 4, 5 etc. It's very orderly. to get a segment of the train, you use curly brackets {}. {0} would return whatever number is in the first box of RAM. as quigibo said, do not overwrite this data, as the OS uses it, and you could easily screw up your calculator by just filling random addresses with numbers. However, quigibo has "safe RAM" areas in axe, which you can modify as you wish, as DJ said, they're L1-L6. L1-L6 are called "pointers". i personally like to call them in my head as "markers", since to me they mark segments of the train. each safe RAM segment contains a certain number of boxes. so 5->{L1} stores 5 into the box at L1. now i've been using the term "box" to describe each segment on the train, aka each segment of RAM. for the record, each segment of RAM is 1 byte. i won't get technical, but basically that means you can store between 0 and 255 into each box. if you try 256->{L1}, the numbers loop back around. so L1 will hold the value 0. 0=256. 1=257. i know. crazy. however, variables A-Z+theta can hold up to 65535.

variables have always intrigued me at first, as i never understood them in Axe. one more long paragraph:

A-Z+Theta are pointers. they point to a box number. 0->A does not store the value 0 into A. it tells A to "point" to RAM address 0. remember how L1-L6 are just "markers" on the train of RAM? same thing with variables. basically, you should rarely use curly brackets {} around just a variable, as that will usually turn out to be some information that the OS uses to run, and if overwritten, may RAM clear. basically, variables are like markers of RAM addresses.

We're all familiar with TI-Basic. I think there should be a similar language for OT calculators. OT basic is not a final name. That can be discussed later. However, if you agree there should be a basic-like language, what are commands and features you would like to see? Named variables? Compiled rather than interpreted? I want radical ideas, things you wish you could do. However, I have one condition. Please no features that are gaming specific!! This is for math programs mainly. No requests for sprite routines or the like obviously targeted for gaming.  Command suggestions will be edited in regardless of feasibility.

Note: This topic should have extremely little discussion. Do not shoot down anyone's idea. Don't speculate on whether hardware can handle it or not. We can discuss fine details after the calculators hardware is planned. I ask this so this thread is as uncluttered as possible! Thank you.

Data Structures and Variable Types:
-Lists
-Matrices
-Multidimensional arrays
-Strings
-Character arrays
-Declared Variables
-Int, Float, Bool etc. types
-Prefixed variables to define their types. (ex. sVAR would be a string)
-Vectors


Variable Commands:
-ToUpperCase/ToLowerCase
-Pic recalling/storing
-Undeclare/Delvar function to free up memory
-Conversions between data types (i.e. list>Matrix, String>Number, Number>String etc.)
-Sum/Mean/Median of a List/Matrix/Array
-Random numbers, integers, matrices, lists, other data types
-Advanced Array functions

Graphical Commands:
-Pxl-On/Off/Change
-Pt-On/Off/Change
-Line
-Circle
-Polygon
-Graph Function
-Shade
-Clear Screen
-3-D support and polygons
-Operating System GUI elements, like DCS libs.

Control Statements:
-If (File/Variable Exists)-Else-End
-If-Else-End
-For
-While
-Break
-Try/Catch
-Goto/Lbl

I/O Commands:
-Getkey (multiple keypress getkey)
-Link Send
-Link Receive
-Prompt/Input
-Disp/Output/Print

Operators:
-Multiplicative
-Subtraction
-Negativity
-Addition
-Division
-Modulus
-Assignment
-Equality
-Inequality operators (greater than, less than etc.)
-Exponent
-Logical not()
-Logical and
-Logical or
-Logical xor
-Increment command (+= in C)
-Decrement commmand (-= in C) same with multiply, divide, modulus.


Math Functions:
-Every math function in the OTz80 calc that can be used outside a program on the "homescreen"
-include sorting lists


Language Design:
-Two languages, one interpreted, one compiled?
-Syntax highlighting! (great suggestion willrandship)
-Search function (ctrl+f)
-Numbered lines in the GUI
-SkipTo command, to skip to a programmer-defined place in the code, or waypoints like 1/4, 1/2, 3/4 through the code.
-Descriptive error messages when a program errors.
-Key to break execution of programs (like [On])
-Copy/Paste sections of code
-Undo/Redo function.
-Token based/Non-token based, still needs to be decided.
-Access to foreign characters and lowercase.

i never understood why people are opposed to dvorak. for example, this post was typed using dvorak. regardless, ABCDE should be fine as SAT/ACT regulators are used to that kind of layout.