aixtal has asked for the wisdom of the Perl Monks concerning the following question:

Dear Monks,

I just found out that anonymous hashes do not have a stable address throughout program execution : 

use Scalar::Util qw(refaddr);

for (1 .. 10) {

    print refaddr { foo => 1 }, "\n";
    
}

[download]
The above fragment of code prints out adresses that seem to run in a cycle of 3 : 
4303377064
4303375648
4303377184
4303377064
4303375648
4303377184
4303377064
4303375648
4303377184
4303377064

I guess this is due to some compilation internals. But still, I found it puzzling. I was naively expecting an anonymous hash to be compiled once and for all at a given place.

I wonder if some monk has an insight on this ?

Replies are listed 'Best First'.
Re: Address of anonymous hash
by Corion (Patriarch) on Feb 02, 2011 at 07:47 UTC

    It's exactly the reverse. An anonymous hash needs to be created every time you assign it, because the only thing you can store is a reference to it. The difference shows in the two programs:

    #!perl -w
use strict;
use Data::Dumper;
my @hashes;
for (1..2) {
    push @hashes, { foo => 'bar' };
};
$hashes[0]->{baz} = 1;
print Dumper $hashes[1]; # { foo => 'bar' }

    [download]

    Your expectation would change this to the following:

    <c>
#!perl -w
use strict;
use Data::Dumper;
my @hashes;
my $constant_hash = { foo => 'bar' };
for (1..2) {
    push @hashes, $constant_hash;
};
$hashes[0]->{baz} = 1;
print Dumper $hashes[1]; # { foo => 'bar', baz => 1 }

    [download]
[reply]
[d/l]
[select]
Re: Address of anonymous hash
by davido (Cardinal) on Feb 02, 2011 at 08:11 UTC

    Your example creates a new anonymous hash that falls out of scope with each iteration of the loop. So in the simplest terms, it's not the same anonymous hash each time. The fact that a newly created anonymous hash has or doesn't have the same address as a previous one that has fallen out of scope is practically coincidental. More importantly, it shouldn't matter what is happening behind the curtain.


    Dave

[reply]
      Understood! Many thanks to Corion and you, these explanations are crystal clear !
[reply]
Re: Address of anonymous hash
by ikegami (Patriarch) on Feb 02, 2011 at 17:35 UTC
    You can think of «{ ... }» as «Hash->new(...)». (Of course, what it returns isn't blessed.)
[reply]
[d/l]
[select]
Re: Address of anonymous hash
by locked_user sundialsvc4 (Abbot) on Feb 02, 2011 at 15:30 UTC

    You should never use a memory-address as a “key.”   It is never wise to build code that is in any way dependent upon “an implementation side-effect,” even if that side-effect appears to be stable.   You will get burned ...

      You should never use a memory-address as a “key.”

      Have you heard of inside-out objects? At the base, they do exactly what just proclaimed shouldn't be done. 

      {
   my %objs;

   sub new {
      my $obj = { ... };
      my $key = 0+$obj;
      $data{$key} = $rec;
      return $key;
   }
}

      [download]

      (Blessing omitted for clarity.)

      I don't know to what side-effect you are referring. That numification of a reference returns its address? That's safe.

[reply]
[d/l]

      You should never use a memory-address as a “key.”

      Have you heard of inside-out objects? At the base, they do exactly what just proclaimed shouldn't be done. 

      {
   my %objs;

   sub new {
      my $obj = { ... };
      my $key = 0+$obj;
      $data{$key} = $rec;
      return $key;
   }
}

      [download]

      (Blessing omitted for clarity.)

      I don't know to what side-effect you are referring. That numification of a reference returns its address? That's safe.

[reply]
[d/l]
      A reply falls below the community's threshold of quality. You may see it by logging in.

Back to Seekers of Perl Wisdom