Re: Power of two round up.

[I0]

local $_ = (shift)-1;
$_ |= $_>>1; $_ |= $_>>2; $_ |= $_>>4; $_ |= $_>>8; $_ |= $_>>16;
return ++$_

[Dominus]

I'm amazed at how clever this is, and I'm going to use it in my program. Thanks a lot.

[I0]

Although it's a technique more appropriate in a C program. In Perl, log($_)/log(2) benchmarks faster. (assuming it rounds properly at the boundaries)
Update: My earlier timethese qq{} may have been misleading. With timethese sub{} the | >> method comes out faster

[Dominus]

Sure, because it has to dispatch many fewer opcodes. But if I had wanted to use the log₂ x solution, I wouldn't have asked in the first place. Floating-point arithmetic gives me the heebie-jeebies.

[runrig]

Just to give Dominus more heebie jeebies :)

sub round_up {
 local $_ = (shift)-1;
 my $num = (2**int(log($_)/log(2)))-1;
 return ++($_ |= $num);
}

[jynx]

On the other hand,

If you want to use horrible and/or (somewhat) obfuscated code, you could try:

sub shifty {
   my $v = (shift)-1;
   map { return ++$v } map { $v |= $v >> $_ } (1,2,4,8,16);
}

or alternately (the readable version):

sub shifty2 {
   my $v = (shift)-1;
   $v |= $v >> $_ foreach (1,2,4,8,16);
   return ++$v;
}

i don't know why i was compelled to post that, but i knew one could reduce the repition and possibly obfuscate the code a little with some obviously useless uses of map doing things it wasn't meant to do...
my $0.02;

jynx

[I0]

#or perhaps
$w=32; $v |= $v>>($w>>=1) while $w;