http://www.perlmonks.org?node_id=668389


in reply to Reference assessment techniques and how they fail

Note that this technique can't detect a code reference because it would call the referenced sub.

It calls the overloaded sub, which has to return a code reference to execute. If you write defined &$some_object, perl will not call the returned code reference, but the expression will evaluate to true.

Replies are listed 'Best First'.
Re^2: Reference assessment techniques and how they fail
by kyle (Abbot) on Feb 17, 2008 at 17:18 UTC

    Good point, but there's actually another problem I forgot to mention. In a &{}, undef is code.

    use strict; use warnings; use Test::More 'tests' => 3; sub is_code { no warnings qw( void uninitialized ); return eval { defined &{$_[0]}; 1 }; } sub real_sub { die 'real sub called' } my $sub_ref = sub { die 'sub ref called' }; my $undef; ok( is_code( \&real_sub ), 'real sub ref is code' ); ok( is_code( $sub_ref ), 'lexical sub ref is code' ); ok( is_code( $undef ), 'undef is code' );

    This isn't a big deal, I guess. We just have to change is_code to return defined $_[0] && eval { defined &{$_[0]}; 1 };. Still, I'd rather call something that figures out what it is than call it to figure out what it is.

      Good point, but there's actually another problem I forgot to mention. In a &{}, undef is code.

      That's not true. Why is your eval always returning 1 unless it dies? With

      sub is_code { no warnings qw( void uninitialized ); return eval { defined &{$_[0]} }; }

      things look more reasonable. If undef were code, it'd be a serious bug.

      --shmem

      _($_=" "x(1<<5)."?\n".q·/)Oo.  G°\        /
                                    /\_¯/(q    /
      ----------------------------  \__(m.====·.(_("always off the crowd"))."·
      ");sub _{s./.($e="'Itrs `mnsgdq Gdbj O`qkdq")=~y/"-y/#-z/;$e.e && print}

        Ah, you're right. I was thinking of the wrong exception and wrote the wrong test.

        use Test::More 'tests' => 5; sub is_code { no warnings qw( void uninitialized ); return eval { defined &{$_[0]} }; } sub real_sub { die 'real sub called' } my $sub_ref = sub { die 'sub ref called' }; my $undef; ok( is_code( \&real_sub ), 'real sub ref is code' ); ok( is_code( $sub_ref ), 'lexical sub ref is code' ); ok( ! is_code( $undef ), 'undef is not code' ); ok( ! is_code( 'string' ), 'string is not code' ); ok( is_code( 'real_sub' ), 'string is code' );

        So you still need an extra check in is_code.

        sub is_code { no warnings qw( void uninitialized ); return '' ne ref $_[0] && eval { defined &{$_[0]} }; }

        That's fairly simple, but I think I still prefer the battery of ref, Scalar::Util::reftype, Scalar::Util::blessed and overload::Method.