Tue Jan 30 21:09:25 GMT 2007
instead of rewriting the core interpreter bottom up, it might be a
good idea to keep the core as it is.. i went through a couple of cuts
that would have benefitted from a bit more thought..
i'm not sure wether incremental rework actually takes more
time.. maybe bug wize there's a gain. however, it does require a bit
so. towards a rewritten interpreter, following the lit/prim/word
approach with return stack bit.
-> eliminate most instruction lookahead: replace with a single
-> eliminate most return stack tricks (not necessary?)
-> separate out dictionary and error handling (error = packet?)
that's probably enough already. woah. can of worms :)