Re^3: possible combinations in sequence

[roboticus]

ikegami:

Thanks for the tip on the speed boost. Once I get the benchmarking stuff installed I'll play with it. While I do like your suggestion, I prefer the order that my method generates--all one-word combinations first, then the two-word combinations, etc.)

Re: shoddy code. Yeah, I guess so, consider me properly admonished. ++ for calling me on it and keeping me honest. When I thunk up the technique to use, I just erased the original function body and started whacking away at it. I didn't think to clarify things by using better variable names. (Of course, I just came off of a little golfing trip so my head was in "trim keystrokes" mode.</lame_excuse_mode>) Now, I guess the proper thing to do is to clean it up a little and insert your suggestion, so here goes:

<pedagogical_mode>

sub function {
    my @parts = split /:/, shift;

    # Null is the complete list of combinations for
    # an empty word list
    my @combinations=();

    # Sequentially (recursively with tail recursion
    # removed) rebuild the combination list adding one
    # new word each iteration
    for my $new_word (@parts) {

        # Given a complete set of combinations for a
        # given list of words, we can add a new word to
        # the list and generate all new valid combinations
        # by concatenating to the original list:
        push @combinations,

              # the new word (a single word is a valid
              # combination)
              $new_word,

              # and the original list with the new word
              # glommed onto the end of each member
              map {$_.':'.$new_word} @combinations
              ;
    }
    return @combinations;
}

</pedagogical_mode>

--roboticus