Re: in praise of tabling
Maori Scarey wrote:
Yes but I guess the subtle parts of the tabling strategy is to
effectively table calls, not returns, so as to act on recursion loops.
No. Or maybe "inclusive no" if that makes sense.
Tabling calls is one thing, necessary, but not sufficient, to do decent
tabling. Tabling just calls can prevent some loops (the ones that
tabling or memoing in general can catch). Tabling "returns" is the hard
part, because one would like to share as much of the return (the answer)
with the call (and other calls) as possible: that's the thing that makes
the data structures for doing tabling "subtle" - as opposed to doing
assert/retract and do-it-yourself-memoing.
Cheers
Bart Demoen
.
