michaelmrose / assembler.clj Goto Github PK

Specifically the 28th line which ought to be the result of parsing @KEYB0ARD_NOT_PRESSED or 35 in binary 100011 ends up being 20 in binary 10100.

symbolic addresses that don't resolve to labels are assigned to addresses starting from 16. The last one created is in fact 19 so 20 would make sense if @KEYBOARD_NOT_PRESSED wasn't resolved but it should have been added to the symbol-map atom in the prior step by build-label-map-and-preprocess and indeed inspection shows it is there.

Debugging transform-nodes with cider causes it to emit the correct instruction which is interesting.