TL;DR

I’m still intrigued by Raku, though not completely sure about it.

I’ve slowly started to take a look at the Advent of Code, 2018 edition, and I hit a small problem that I thought would be perfect to translate into Raku.

I mean, so far I did not read file inputs, nor use too many data structures. So why not?

I will not spoiler the fun if you’re into it, just share the code and a couple notes. It’s day 3 if you’re curious.

#!/usr/bin/env raku
use v6;

sub MAIN ($input = '03.tmp') {
   my $fh = $input.IO.open;
   my %fabric;
   my %double is SetHash;
   my %free is SetHash;
   for $fh.lines -> $line {
      my ($id, $left, $top, $width, $height) = $line.comb(/\d+/);
      %free.set($id);
      for +$left ..^ $left + $width -> $h {
         for +$top ..^ $top + $height -> $v {
            my $key = "$h:$v";
            if (%fabric{$key}:exists) {
               %double.set($key);
               %free.unset(($id, %fabric{$key}));
            }
            else {
               %fabric{$key} = $id;
            }
         }
      }
   }
   say %double.elems;
   put %free.keys;
}

The implementation comes out compact, but in hindsight I first solved it in Perl, so I had all inputs when I moved on to Raku, i.e. not only the algorithms but also the knowledge on how to merge them.

I was badly hit by the fact that the string extraction in the ... = $line.comb(...) line returns all strings. This did not always produce the expected looping in the two nested for later on, and made me force the interpretation as integers with a + (as in +$left and +$top). I have to admit that I don’t like this particularly: it seems unPerlish and a loss in whipuptitude.

Update Reddit to the rescue - thanks to [b2gills][] I could have forced the comb into integers right off the bat, with +«$line.comb(/\d+/). It still requires us to do this operation, but at this point it would be sort of complaining about strict…

As a Perl hacker, I’m used to play with scalars, arrays and hashes, covering 99% of my data structure needs. In this case I decided to give SetHash a try; it’s maybe an improvement in readability, although not faster to write. It’s still good that I can do everything the old way with hashes anyway.

I’m still baffled by the sigil invariance, and forget to use %fabric{$key} instead of the perlish $fabric{$key}. No big deal though, the error reporting system is amazing.

One last thing I did not understand is the need to put two sets of round parentheses in this line:

%free.unset(($id, %fabric{$key}));

Withouth them… it was complaining with some obscure error message:

Too many positionals passed; expected 2 arguments but got 3
  in block  at 03.raku line 13
  in sub MAIN at 03.raku line 9
  in block <unit> at 03.raku line 1

Go figure…