Tue Apr 8 13:41:28 EDT 2008

relative addressing

so.. the relative addressing is a bit of a hack. is it possible to
move address resolution down to the assembler opcodes? sure. just have
them depend on 'pointer-get'.

let's port the pic18 assembler, and see if the generation can be
improved a bit.

porting asmgen and trying to get relative addressing, which now has
already overflow detection, to use absolute input.

Maybe meta-catch-undefined can be eliminated by setting undefined
words to 'here', so they compile to a small relative jump
instruction. Maybe just leave that out: it's an optimization, not

OK. pic18 seems to work too.