...as the title says, I have questions about subroutines, arrays, and screen manipulation.
This is in prep for a very secret but awesome project!

Let's get started!

Subroutines
1) In a subroutine, are the letter variables (like A-Z, theta, etc.) isolated from the "main" program?
2) In a subroutine, if I'm not using rn (n is a number) for arguments, can I use them and not worry about crashing?
3) Can I call a subroutine from a subroutine? If so, will I lose my rn variables from the origin called subroutine?
4) Can I get more arguments than just r0-r6, or am I stuck with them?

Arrays
1) How do I make string or number arrays in Axe? Please include as many ways as possible to manipulate the array, and obviously read it!
2) Is it possible to do an array of an array? If so, how?

Screen Manipulation
1) How can I capture the whole or part of the screen? (And restore it?)
2) With that capture, can I use Pt-On( to move it? If not, how would I do so?

BONUS QUESTION - Axioms
1) If I'm right.... if I want to use more than one Axiom, I could flip flop the "includes" (#Axiom) and use the functions interchangably? (Not at the same time tho)
2) Does compiled Axe code need the Axiom appvar or is the appvar's code included inside the compiled prgm?

I might have more questions in the future, so beware!

Thanks in advance!

if you want to capture and move a portion of the buffer around the screen then you are going to need to set up an array in free RAM. find the position of the portion you want(it has to be byte alligned, unless you want to use pixel test) in L6 (the position will be the horizontal, byte-alligned position [0-11] and the vertical, pixel alligned position [0-63]) and do something like this:
[offset to selection in L6]->A
for(N,0,[selection height in bytes-1]
for(M,0,7
for(L,0,[selection width in bytes-1]
{N*8+M*[selection width in bytes-1]+L+A+L6}->{N*[selection width in bytes-1](L*8)+M+L1}
End
End
End
the data for your selection(as long as it's small enough to fit in the free RAM without overflowing. if you need more space than is available in L1, your best bet is to either relocate the vars to a different spot in free RAM, or to create a 768 byte appvar and store it there.) should now be stored in L1. to draw it on the screen using Pt-on simply do something like this:
[X draw offset]->A
[Y draw offset]->B
for(M,0,[selection height in bytes-1]
for(L,0,[selection width in bytes-1]
Pt-On(L*8,M*8,{M*[selection width in bytes-1]*8+(L*8)+L1
End
End
disclaimer:there are probably horrific errors in this code that i am somehow missing, as that tends to be the case. also, these routines can be optimised considerably. they are in their current form for readibility.

Flip-flopping axioms won't work because the axioms are looked at in the order they are imported, and the limit is 5 axioms per program.  If you need to use 2 sets of Axioms, just make sure they use different token combinations and you won't have any conflicts.  If they do use the same token combinations, you can probably change the source and recompile the Axiom to use a different token since I assume most axioms will be open source.

You do not need the Axiom appvar to execute the program, just to compile.  Axioms work a lot like the Asm() command, but smarter and with readable syntax.