NOTE: The behaviour of a "my" statement modified with a statement modifier conditional or loop construct (e.g. "my $x if ...") is undefined. The value of the "my" variable may be "undef", any previously assigned value, or possibly anything else. Don't rely on it. Future versions of perl might do something different from the version of perl you try it out on. Here be dragons.

(emphasis partially mine)

So my question is that why @array is behaving like a static variable in C language here?

#!/usr/bin/perl

use 5.010;


use strict;
use warnings;

sub foo {
    my $x if 0;
    say ++$x;
}

sub bar {
    my $x if 0;
    say ++$x;
    bar() if @_;
}

foo;
foo;
bar(1);

__END__
Deprecated use of my() in false conditional at ./x line 10.
Deprecated use of my() in false conditional at ./x line 15.
1
2
1
1
[download]

Since 5.10, there is a warning on my $foo if 0; (at compile time). People have argued there should always be a warning on any my $foo if EXPR or my $foo = EXPR1 if EXPR2; usage, but that hasn't happened, and isn't happening in blead either.

Also since 5.10, if you want a static variable, you can use the state keyword.

Thanks JavaFan and LanX.
Points on compile-time and run-time difference are well taken.
BTW, in my original code, I do use strict/warning, and knew that separating the declaration and condition statement makes it work as expected. But still knowing a little more about what is going on under the hood is more satisfactory.

2009-05-07	perl5ever	scope of my $x in if statements	SoPW
2009-02-12	sandy1028	Difference between my ($foo) & my $foo	SoPW
2008-01-14	parv	"my $p = value if condition()" as bad|deprecated as "my $p if 0" (Naughty indeed)	SoPW
2007-12-29	Limbic~Region	What about if (my $var = foo()) { ... }	SoPW
2005-04-29	mda2	$_ = "My First JAPH"; print ( (split(/\s/, $_), Multi)0,3,2) if ( $,=" " )	Obfu
2005-01-26	szabgab	my $cache = undef if undef;	SoPW
2004-03-01	Roger	Perl Idioms Explained - my ($foo, $bar) = @{shift(@_)}{qw/ -foo -bar /}	Med
2001-07-14	MeowChow	It *is* a feature: mod_perl and 'my $x if 0'	Med

It's quite simple: Using a my variable without having first executed the my is a bug. By braking that rule, you break the "fourth wall" between the promised behaviour of my and its optimised implementation.

my variables are created at compile time. Executing a my places a directive on the stack that will get executed on scope exit. The directive causes the variable to be cleared (if possible) or replaced (if a reference keeps it alive).

If you want a static variable, use state.

Thanks, I forgot about scope-exit behavior...

Anyway I'm wondering why a post-fix if doesn't force a new scope, IMHO that would solve the problem in a consistent way:

use strict;
use warnings;
 
$a=0;
 
if ($a) {
  my $z=666;
}
 
my $y=666 if $a;
 
print $y;         # -> nothing
print $z;         # -> Global symbol "$z" requires explicit package na
+me
[download]

of course treating short circuit and in the same way could cause more complications.

$a and my $x =666;
print $x;        # -> nothing
[download]

At least this "it's a new scope" logic could be used to detect "my"-problems at compile-time and throw warnings.

Cheers Rolf

UPDATE: short-circuit and

It wouldn't fix the following:

foo() and my $x;
foo() or my $x;
foo() && my $x;
foo() || my $x;
foo() // my $x;
foo() ? my $x : 1;
[download]

Not to mention that some code relies on the current behaviour.

sub foo{
        my $var = shift;
        my $href = shift;
        my @array = $href->{pass_in} if defined $href and exists $href
+->{pass_in};
        push @array, $var;

        print "contents : @array\n";
}

foo(1);
foo(2);
[download]

However, commenting out the unused href code and declaring array as a lexical like:

sub foo{
        my $var = shift;
        my @array; 
        #my $href = shift;
        #my @array = $href->{pass_in} if defined $href and exists $hre
+f->{pass_in};
        push @array, $var;

        print "contents : @array\n";
}

foo(1);
foo(2);
[download]

contents : 1
contents : 2

$href is empty, since it is not passed into foo. hence @array ends up being a package variable and not a lexical variable.

As others have pointed out, that's wrong. my creates lexicals at compile time. It doesn't have a our fallback. The 'if' modifier with a false condition inhibits the opcode which clears the lexical at runtime.

our @array; # package vaiable
@array = qw( foo bar );

sub foo{
        my $var = shift;
        my $href = shift;
        my @array = $href->{pass_in} if defined $href and exists $href
+->{pass_in};
        push @array, $var;

        print "contents : @array\n";
}

print "outer scope - contents : @array\n";

foo(1);
foo(2);

print "outer scope - contents : @array\n";

__END__
outer scope - contents : foo bar
contents : 1
contents : 1 2
outer scope - contents : foo bar
[download]

As you can see, the variable @array inside the sub is a lexical.