Benjamin Goldberg wrote in perl.golf : > This might be a bit of a strange idea, but... suppose that for scoring, > instead of using total number of bytes for the score, you used the total > number of opcodes. The B module, or one of the backends invoked through > -MO=<whatever>, would be used for finding out this info.
Interesting idea... with less ops, a compiled perl program takes less memory, and may run faster. Plus, it's a way to learn about the internals. Would ops that have been optimized to nullops counted in the final score ? Also, you'll need to standardize to a specific perl version -- perl 5.8 + ithreads compiles to a different set of opcodes. 'perl -MO=Concise foo.pl | wc -l' should be enough if you don't discard opcodes. (if not, we can hack a quick module B::CountOpcodes or whatever. Duh, it would have to handle subroutines and BEGIN blocks too...) Just, please don't use the namespace B::Golf for it. > You would have to forbid the various ways that code can be hidden from > perl's introspection, but that shouldn't be *too* much of a hardship for > golfers.
