There are several really good solutions at Eek! goto? This was when the accidental duplicate nodes stuff was happening so you probably want to look at both nodes.

BrowserUK shows a really elegant solution using goto, I posted a hash dispatch method, Abigail-II offers discussion..... Switch even gets a mention but this uses 500 lines and goto internally anyway.

cheers

tachyon

s&&rsenoyhcatreve&&&s&n.+t&"$'$`$\"$\&"&ee&&y&srve&&d&&print

Very interesting, especially since BrowserUk's node is the exact same code that I'm looking at! GMTA and all that.

—John

This was my final version of the algorithm. It benchmarked quickest of all the versions (7) that I tried, but I haven't found a simple way of verifying that it's a faithful re-implementation of the C version.

use constant A => 0;
use constant B => 1;
use constant C => 2;

sub mix ($$$) {
    use integer;
    $_[A] -= $_[B]; $_[A] -= $_[C]; $_[A] ^= ($_[C]>>13);
    $_[B] -= $_[C]; $_[B] -= $_[A]; $_[B] ^= ($_[A]<< 8);
    $_[C] -= $_[A]; $_[C] -= $_[B]; $_[C] ^= ($_[B]>>13);
    $_[A] -= $_[B]; $_[A] -= $_[C]; $_[A] ^= ($_[C]>>12);
    $_[B] -= $_[C]; $_[B] -= $_[A]; $_[B] ^= ($_[A]<<16);
    $_[C] -= $_[A]; $_[C] -= $_[B]; $_[C] ^= ($_[B]>> 5);
    $_[A] -= $_[B]; $_[A] -= $_[C]; $_[A] ^= ($_[C]>> 3);
    $_[B] -= $_[C]; $_[B] -= $_[A]; $_[B] ^= ($_[A]<<10);
    $_[C] -= $_[A]; $_[C] -= $_[B]; $_[C] ^= ($_[B]>>15);
}

use constant GOLDEN_RATIO    => 0x9e3779b9;
use constant INITHASH        => 2;

sub hash {
    use integer;
    my($a,$b,$c) = (GOLDEN_RATIO, GOLDEN_RATIO, $_[INITHASH]);
    my ($p, $len) = (0, length $_[KEY]);
    do {
        my($x,$y,$z) = unpack 'V3', substr($_[KEY] . (chr(0)x11), $p, 
+12);
        mix($a += $x, $b += $y, $c += $z + $len);
        $p+=12;
    } while $p <= $len;
    return $c;
}
[download]

Hm? Unless I'm missing something, that leaves out the part that's relevant to this thread of how to best code the case fall-through thing in Perl.

If the input is not a multiple of 12 bytes, it needs to do this (from the original C version):

   switch(len)              /* all the case statements fall through */
   {
   case 11: c+=((ub4)k[10]<<24);
   case 10: c+=((ub4)k[9]<<16);
   case 9 : c+=((ub4)k[8]<<8);
      /* the first byte of c is reserved for the length */
   case 8 : b+=((ub4)k[7]<<24);
   case 7 : b+=((ub4)k[6]<<16);
   case 6 : b+=((ub4)k[5]<<8);
   case 5 : b+=k[4];
   case 4 : a+=((ub4)k[3]<<24);
   case 3 : a+=((ub4)k[2]<<16);
   case 2 : a+=((ub4)k[1]<<8);
   case 1 : a+=k[0];
     /* case 0: nothing left to add */
   }
   mix(a,b,c);
[download]

And you left out the $c += $length (using 32-bit integer overflow).

Hmm again... or is the . (chr(0)x11), $p, 12 taking the place of that, to fill out the last "gulp"?

So even if that's not quite right, your answer to how to write such a switch statement is "don't. a more natural way (at a higher level in the algorithm) in Perl suggests itself and prevents the need".

Back to the other thread: thanks for sharing fastest implementation of the mix() function you found in pure Perl.

    case  FOO:
    case  BAR: /* code here */
               break;
[download]

That is, having more values map to the same code.

How to do that efficiently in Perl? It depends. Sometimes a computed goto is the best way, sometimes a series of if statements. Both have the disadvantage they potentially spend a lot of time finding the code that needs to be executed. And depending on the values of the cases, and in which order they are, you could use an array of code references, and execute each code reference from the switched value to the end.

Abigail