Notes on assembling things for IC

Notes on assembling things for IC:



Add as11_ic stuff to your path:
	set path= ( $path ~/mac/robots/as11 )
Run as11_ic file.asm to assemble a file and produce an .icb output.
Run as11_lis file.asm to assemble a file and produce an .icb output,
    plus generate a .lis file displaying the output of the assembly.

To generate a new pcode.s19, do the following on the appropriate version of pcode.asm (i.e. the filename might be different than in the example):

	# On zealand or another box that mounts CS department home dirs
	set path= ( $path ~/mac/robots/as11 )
	cd ~jonh/robots.bak/math/pcode/pcode-2.81-dist/pcode
	as11_cpp pcode.asm -DBOOKBOT -DLCD_ROWS=2 -DLCD_COLS=16 -I../libs > pcode.out

As of this writing, the most recent pcode asm was ~jonh/robots.sk/pcode_rwmem.asm. It includes the FDIV fix, the flt2int hack that turns underflow errors into 0.0, and the pacman status indicator, and it omits the bogus check for DIGINPUT, which the Rug Warrior doesn't have. (DIGINPUT is at 0x7000 on some pcode boards, but we put code there, which confused that old check.)

You can get a listing of the pcode assembly process this way, but you'll get a useless .icb instead of an .s19:

	as11_lis pcode.asm -I../libs > pcode.out

Could we download the icc11 project from cher, compile a floating divide, and use that? Requires a PC.

Another concern we have is precious bytes of RAM. Possible sources to reclaim bytes:

There are a few arrays & routines I've allocated for debugging. Naturally, the first step should be to walk through sources looking for dead code & data.
I have pcode source -- maybe it sloppily allocates stacks & things; we may need fewer than it allocates.
pcode links to math11routines.asm, which it doesn't seem to use all of. I doubt the linker is smart, so maybe we can strip down math11routines.asm.
If we store a lot of SE(2) matrices, we should consider reducing to [theta x y]. (That works, right?) Costs a little time, but...
The math11routines.asm stuff does arithmetic in degrees; then pcode converts to radians. But some math routines (ATAN) actually work in radians and convert to degrees. We could remove the rad->deg->rad conversion stuff.