There are a few problems. One is that %s{'a'} = 10; won't compile. You mean to say, $s{'a'} = 10;

The next problem is that testa() returns a flat list of keys and values, not a hash. You could re-hashify them like this:

%{ { testa() } }
[download]

This takes the return value of testa() (a flat list of keys interlaced with values), absorbs that list into an anonymous hash (the inner set of {...} does this), and then dereferences that anonymous hash ref (the outter %{ ..... } does this).

But even then you still have a problem. Your loop will print a list of keys, sure enough. But how are you going to access the values if nothing is holding onto that hash ref?

Try this instead:

while( my( $key, $val ) = each( %{ { testa() } } ) {
    print "Key: $key\tValue: $val\n";
}
[download]

...or better yet, pass by reference, as described in perlsub:

while( my( $key, $val ) = each( %{ testa() } ) {
    print "Key: $key\tValue: $val\n";
}
sub testa {
    my %s = (
        a => 10,
        b => 20,
        c => 30
    );
    return \%s;
}
[download]

sorry for all the confusion about the typo.

But thnx for the %{ { &testa } } that does the trick!
I guess it might be very usefull to study the perlsub (how do you guys make this a link, where is it documented, I've search the docs already a couple of times ?)!!

LuCa

You're welcome, but I did say %{ { testa() } }, not %{ { &testa } }. Your specific example code doesn't bump into the difference, but there is a difference. It's just good to be in the habit of using the subroutine() method of calling subs, instead of the Perl4-ish (and sometimes confusing) &subroutine method.

As for how to link to perlsub, see About the PerlMonks FAQ, and specifically, What shortcuts can I use for linking to other information?. ...but first read perlsub; that's a much more important step toward your understanding Perl.




Dave

%s{'a'} = 10 ;
%s{'b'} = 20 ;
%s{'c'} = 30 ;
[download]

$s{'a'} = 10 ;
$s{'b'} = 20 ;
$s{'c'} = 30 ;
[download]

#!/usr/bin/perl
use strict  ;
use warnings ;


foreach ( keys %{&testa} ) {
    print "key is $_\n" ;
}

sub testa {
    my $s = () ;
    $s->{'a'} = 10 ;
    $s->{'b'} = 20 ;
    $s->{'c'} = 30 ;

    return $s ;
}
[download]

Note that:

    my $s = ();
[download]

is equivelent to

    my $s = (undef);
[download]

which is equivelent to:

    my $s;
[download]

The first use of a hash ref autovivifies the referred hash making the previous assignment redundant (and misleading). Consider:

use strict;
use warnings;
use Data::Dump::Streamer;

my $s = ();
Dump ($s);
$s->{'a'} = 10 ;
Dump ($s);
[download]

which prints:

$VAR1 = undef;
$HASH1 = { a => 10 };
[download]

and note that the contents of $s changes from an undefined value to a hash ref.

If you really want to indicate intent with an assignment use:

    my $s = {};
[download]

---

DWIM is Perl's answer to Gödel

nope, the sub should return a hash and not a hash reference. It should be the same as with

foreach( keys CGI::Vars() )
[download]

because that is what I'm trying to do!

LuCa

You can't "return a hash" from a subroutine, you can only return a list, the effect of return %hash is to flatten the hash into a list of key => value pairs. Thus the effect of the the three following subroutines is similar (the order of the hash keys in the first aside:)

sub foo
{
    my %foo = ( Foo => '1',
                Bar => '2');
    return %foo;
}
                                                                      
+          
sub bar
{
   my @bar = qw(Foo 1 Bar 2);
   return @bar;
}
                                                                      
+          
sub baz
{
   return 'Foo','1','Bar','2';
}
                                                                      
+          
print foo(),"\n";
print bar(),"\n";
print baz(),"\n";
[download]

This is why people are telling you to return a hash reference as this is the only way to retain the, er, hashiness after the appropriate dereferencing.

/J\

Please show working code which does what you're trying to do. When I try the following, it doesn't work:

Q:\>perl -MCGI -le "foreach( keys CGI::Vars() ) { print }"
Type of arg 1 to keys must be hash (not subroutine entry) at -e line 1
+, near ") ) "
Execution of -e aborted due to compilation errors.
[download]

When I try the following, which the documentation and davidj suggest, it works (in the sense of not being a syntax error). But that works by returning a hash reference:

Q:\>perl -MCGI -le "foreach( keys %{ CGI::Vars() }) { print }"

Q:\>
[download]