Mon Jun 25 16:02:53 CEST 2007

literals : save

oops. too much coffee, going too fast..  i AM doing SAVE for each
literal, so the postprocess step shoul maybe perform the save too?
this is a bit more complicated than i thought..

so: when to do SAVE?

currently save does:

 ((['drop] save)            '())
 (([op 'POSTDEC0 0 0] save) `([,op INDF0 1 0]))
 ((save)                    '([dup]))

what about just a second compiler pass with the word 'save' ?

that 2nd pass seems to work. problem now is a lot of the literal
macros do need their arguments. am i going to try to fix that now?
maybe think a bit about how to do that in a smart way..

ok. roadmap:

- just added c> == '(qw) op>asm
- replace all lit macros by a qw macro, and remove them from expose hack

done, now for the calls

- replace branches and calls with pseudo ops
- fix vm control ops
- start working on the control part of synth