#!/usr/bin/perl

use strict;
use warnings;

my (@Cdevices,@Mdevices);

push (@Cdevices, 'UnitID', 'fileno');
push (@Mdevices, \@Cdevices);

# this will show us the actual layout of the array
use Data::Dumper;
print Dumper(\@Mdevices);

# $k=0; $k<=$#Mdevices; $k++ is C-style. Possible, but unnecessary ove
+rhead
for my $ref (@Mdevices) {
    print join(' - ', @{$ref}),"\n";

}
[download]

$VAR1 = [
          [
            'UnitID',
            'fileno'
          ]
        ];
UnitID - fileno
[download]

my (@Cdevices,@Mdevices);

push (@Cdevices, 'UnitID1', 'fileno1');
push (@Cdevices, 'UnitID2', 'fileno2'); # <- @Cdevices is now qw(UnitI
+D1 fileno1 UnitID2 fileno2)
push (@Cdevices, 'UnitID3', 'fileno3'); # <- and so on
push (@Mdevices, \@Cdevices);

# this makes it clear:
use Data::Dumper;
print Dumper(\@Cdevices);
print Dumper(\@Mdevices);

for my $ref (@Mdevices) {
    print $ref->[0],"\n";
}
[download]

Indeed. The fragment offered can be fixed...

push (@Mdevices, [$UnitID, $fileno]);
for ($k=0; $k<=$#Mdevices; $k++)
  {
    print $Mdevices[$k][0],"\n";
  }
  $k=0;
[download]

...the named @Cdevices is not required.

If the more complete code is along the lines:

  my @Mdevices ;
  my @Cdevices ;
  while (... something ...) {
    ... do something to get $UnitID & $fileno ...

    push (@Cdevices, $UnitID, $fileno);

    ... do some other stuff ...

    push (@Mdevices, \@Cdevices);
  } ;
[download]

then @Mdevices is going to end up with 'n' copies of a reference to @Cdevices, which contains a list of the 'n' $UnitID and $fileno seen.

Whereas the almost identical:

  my @Mdevices ;
  while (... something ...) {
    my @Cdevices ;
    ... do something to get $UnitID & $fileno ...

    push (@Cdevices, $UnitID, $fileno);

    ... do some other stuff ...

    push (@Mdevices, \@Cdevices);
  } ;
[download]

fills @Mdevices with 'n' references to 'n' different annonymous arrays, each containing a $UnitID and $fileno pair.

Obvious, really.

Alternatively:

  my @Mdevices ;
  while (... something ...) {
    ... do something to get $UnitID & $fileno ...

    ... do some other stuff ...

    push (@Mdevices, [$UnitID, $fileno]);
  } ;
[download]

will do the trick -- unless the @Cdevices array is required in the loop.

Hello!

This code really did the trick!


 my @Mdevices ;
  while (... something ...) {
    ... do something to get $UnitID & $fileno ...

    ... do some other stuff ...

    push (@Mdevices, [$UnitID, $fileno]);
  } ;
[download]

Thanks to all of you for help

Igor

use Data::Dumper;
print Dumper(\@Cdevices);
print Dumper(\@Mdevices);
[download]

This may also be of use: References quick reference

you should de-reference using e.g. print $Mdevices[$k]->[0]

$Mdevices[$k]->[0] is functionally the same as $Mdevices[$k][0] (what the OP had).  You only need the arrow if it's at the 'first' level (i.e. $Mdevices->[0] is not the same as $Mdevices[0] — here, the parser needs some way to disambiguate). In case it isn't ambiguous (between brackets/braces), the arrow is optional.

TFT almut, the old saying about a day learning nothing is a day lost just came home to roost - that's another tool for my toolbox.

It also underlines my sig, doncha think ?

A user level that continues to overstate my experience :-))

oshalla has addressed your coding error, but if you intend to further your Perl programming skills there are a couple of other things you did "wrong".

First, always use strictures (use strict; use warnings;).

Write Perl not C. In particular, Perl for loops are much safer and clearer than C for loops. Consider:

for ($k=0; $k<=$#Mdevices; $k++)
  {
    print $Mdevices[$k][0],"\n";
  }

for my $device (@Mdevices) {
    print $device->[0], "\n";
}
[download]

In fact a slightly more experienced Perl programmer would use a for statement modifier and the loop becomes:

print $_->[0], "\n" for @Mdevices;
[download]

In PerlMonks terms what you did wrong was to not show a runnable sample the demonstrates the problem. The code you provided doesn't show the problem that oshalla (using his elite esp skills) resolved for you. However if you had instead shown us:

for (1 ..5) {
    push (@Cdevices, "UnitId$_", $_);
    push (@Mdevices, \@Cdevices);
}

print $_->[0], "\n" for @Mdevices;
[download]

prints:

UnitId1
UnitId1
UnitId1
UnitId1
UnitId1
[download]

and told us what you expected to see instead we'd have known what the problem was straight off.