• It's a good practice to limit the scope of your variables. It looks like $first, $second and $cycle_length can be declared inside your 1st foreach loop. Similarly, $cycle_length can probably be declared in your sub.
• $number = $number / 2; can be shortened to $number /= 2;. See Assignment Operators.
• perlcritic is a tool which you can use to check for best practices. I ran it on your code, but nothing significant (in my opinion) was reported.

• I would not have made the program ask for input; instead I would read from STDIN. Neither would I print 'Output:' - just print the answer.
• I would not use a bare block in totalizer. Instead I would write: while ($number > 1) {....}. I'd omit the != 0 as well.
• You also may want to do some input validation. You're just assuming everyone will type in exactly two positive numbers; your program is bound to loop forever if someone offers '-1 -1' as input.

$cycle_length being declared globally is not acceptable. I would declare $first and $second in the loop, but they're acceptable where they are.

The other issue it's very hard to spot the loop in your sub. May I recommend

for (;;) {       # I read this as "for ever"
   ...
}
[download]

or the more popular

while (1) {
   ...
}
[download]

I would write this sub like this:

sub totalizer {
    my $number = shift;

    my $cycle_length = 1;

    while($number != 1) {
        $cycle_length++;

        $number = $number % 2  ? $number * 3 + 1
                               : $number / 2;

    }
    return $cycle_length;
}
[download]

Hi the program style looks good. If you are interested in writing your scripts using an editor then you may use Eclipse -- Classic version with Perl Interpreter. It is nice and really good for both error detection, debugging and formatting your code. You may have a look on the editor -- EPIC

This is a more direct link to the problem set on the streamtech site. The link given in the OP requires quite a bit of searching.

I have updated the link in the original post.

Thanks for your replies everyone. Here is what I now have:

#!/usr/bin/perl
use warnings;
use strict;
use 5.010;

# The 3n + 1 problem (http://www.streamtech.nl/problemset/100.html)

my @pairs;
say "Input:";

while (<STDIN>) {
    if (/^[1-9]+\d*\s[1-9]+\d*$/) {    # check for and prevent illegal
+ entries
        chomp;
        push( @pairs, $_ );
    }
    else { say "invalid input!" }
}

say "Output:";

foreach my $pair (@pairs) {
    my ( $first, $second ) = split( / /, $pair );
    my $max_cycle_length = 0;
    if ( $first > $second ) {
        ( $first, $second ) = ( $second, $first );
    }    # doesn't matter whether the higher number is entered first o
+r second

    foreach my $num ( $first .. $second ) {
        my $cycle_length_out = totalizer($num);
        if ( $cycle_length_out > $max_cycle_length ) {
            $max_cycle_length = $cycle_length_out;
        }
    }
    say "$first $second $max_cycle_length";
}

sub totalizer {
    my $number = shift;
    my $current_cycle_length;
    for ( $current_cycle_length = 1 ; $number != 1 ; ++$current_cycle_
+length ) {
        if ( $number % 2 != 0 ) {
            $number = ( $number * 3 + 1 );
        }
        else {
            $number /= 2;
        }
    }
    return $current_cycle_length;
}
[download]

I hope this is an improvement.

Indeed an improvement. For contemplation here's something closer to what I would write:

#!/usr/bin/perl
use warnings;
use strict;
use 5.010;

# The 3n + 1 problem (http://www.streamtech.nl/problemset/100.html)

my @pairs;
say "Input:";

while (defined ($_ = <STDIN>) && !/^\.$/) {
    if (!/^(\d+)\s+(\d+)$/ && $1 > 0 && $2 > 0) {
        chomp;
        say "Invalid input >$_<.";
        say 'Two non-zero positive integers are expected. Input ignore
+d';
        next;
    }

    push @pairs, [$1, $2];
}

say "Output:";

for my $pair (@pairs) {
    my $maxLength = calcCycleLen ($pair->[0]);
    my $length2   = calcCycleLen ($pair->[1]);

    $maxLength = $length2 if $maxLength = $length2;
    say "@$pair $maxLength";
}


sub calcCycleLen {
    my ($number) = @_;
    my $length = 1;

    while ($number > 1) {
        $number = $number % 2 ? $number * 3 + 1 : $number / 2;
        ++$length;
    }

    return $length;
}
[download]

The first change is to add a stopping condition for command line input from STDIN. while (<$fooIn>) { is fine if $fooIn is connected to a file, but is less good for devices such as a keyboard where the end of file isn't clear.

There are a few other changes of note in the while loop:

1. the regular expression is used to capture the two numbers and the requirement for just one white space character is relaxed.
2. A range test is performed on the numbers and provides an additional failure mode for the input
3. The bad input is reported and the valid input criteria are specified
4. There is no else part to the if statement. Instead the "early exit" technique is used. This very often reduces nested code and can make the main flow of the code much easier to see.
5. The pairs are stored in their own two element array. @pairs thus contains arrays - it is an AoA (some people like to use lol instead).

Moving on: as a personal preference I use for everywhere rather than a mixture of for and foreach - probably because I'm lazy.

The @pairs for loop has changed a lot!

1. $pair now is a reference to an array containing two numbers instead of a string.
2. $maxLength is initialised to the first value that will be calculated. This trick is often used where a result is calculated by considering each element of a vector in turn. Initialise the result variable with the calculated value from the first element of the vector, then process the remaining elements in a loop.
3. Calculate the second value. Note the syntax used to access the array elements from the $pair variable holding a reference to the array btw.
4. Update $maxLength if it is smaller than the second length. Note the use of if as a statement modifier.

Now we are in the home stretch.

1. I use my ($number) = @_; because it scales better than a bunch of shifts. Not needed in this case, but it's what flows from my fingers.
2. The C style for loop is almost unused in Perl. In this case a while loop better gives the flavour of what is going on anyway.
3. The ternary operator is easy to abuse. In this case though it reduces some rather cluttered looking code to something that (in my view) is clear and concise.

The final step is to pass Perl::Tidy over the code to clean up any wayward formatting, then enjoy.

True laziness is hard work

Thanks for the link :-) I also tried the first assignment yesterday, but with a small speed-up included. See for comparison:

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

  my %cache = (1 => 0);
  sub _three_n_plus_1 {
    no warnings qw/recursion/;
    my ($n,$count) = @_;
    $count ||= 1;
    return $count + $cache{$n} if exists $cache{$n};
    if($n % 2){
      $cache{$n} = _three_n_plus_1(3*$n+1, $count+1) - $count;
      return $cache{$n} + $count;
    }else{
      $cache{$n} = _three_n_plus_1($n/2, $count+1) - $count;
      return $cache{$n} + $count;
    }
  }

  sub three_n_plus_1 {
    my ($i,$j) = @_;
    my $max = 0;
    for my $x ($i<$j ? $i..$j : $j..$i){
      my $t = _three_n_plus_1($x);
      $max = $t if $t > $max;
    }
    print "$i $j $max\n";
  }
[download]