Okay, I think I can do it:  Select(CONST) will be able to set where you want the variables in an Axe program from that point on in the program.  You can use it multiple times in the same program to swap back and forth between different buffers to effectively increase the number of variables available to the programmer.  The command is a compiler instruction and thus takes no memory in the executable so there is never a penalty to using it.  Also, the variables will from now on be at the end of L1 by default instead of the beginning.  That way, if you do move the variables, you will be able to use the full 768 byte L1 buffer by simply using L1 instead of having to use L1-56.  Future optimizations might also be possible in the future if the variables are kept within that range.

Okay, I think I can do it:  Select(CONST) will be able to set where you want the variables in an Axe program from that point on in the program.  You can use it multiple times in the same program to swap back and forth between different buffers to effectively increase the number of variables available to the programmer.  The command is a compiler instruction and thus takes no memory in the executable so there is never a penalty to using it.  Also, the variables will from now on be at the end of L1 by default instead of the beginning.  That way, if you do move the variables, you will be able to use the full 768 byte L1 buffer by simply using L1 instead of having to use L1-56.  Future optimizations might also be possible in the future if the variables are kept within that range.

EDIT: 1337th post!

Yeah, that would not work.  It would select L2 and then L3 so by the end of that command, it would be at L3.

I might tack on a hash to the front of the token to indicate it is a compiler instruction rather than a language one to make it less confusing.  When I get to adding custom icons, I can also do it the same way:

#Select()
#Icon[]