The whole print vs. printf conversation seems best left for a completely independent thread.

I didn't realise there was any controversy. It's not even a question of print vs printf but one of

my ($h,$m,$s) = ...;
$m = "0$m" if $m < 10;
$s = "0$s" if $s < 10;
my $dur = "$h:$m:$s";
[download]

vs

my ($h,$m,$s) = ...;
my $dur = sprintf("%d:%02d:%02d", $h,$m,$s);
[download]

Right now, you are creating timestamps like 1:3:43 instead of 1:03:43. printf and sprintf are quite apt at formatting this kind of data, but you don't have to use them. Take your pick.

%d    Signed 32/64-bit integer
%2d   ...space padded (on the left) to occupy (at least) 2 columns
%02d  ...zero padded (on the left) to occupy (at least) 2 columns
[download]

hope that you all aren't disappointed that I just used return instead of printf in my subroutine.

No, separation of calculation and I/O is a good thing. However,

print get_duration('s1');
[download]

should be

print get_duration($durations{s1});
[download]

or

print get_duration(@{ $durations{s1} });
[download]

(with the corresponding changes within get_duration). There's no reason for %durations to be global.

Leap seconds? Where have I been? :)

That post tries to show why DateTime wasn't a good solution for this problem, not how to solve the problem using DateTime. Sounds like it worked.

I have read over several perldocs on objects and barely get it

In short,

• Objects represent some object (e.g. a specific user, a specific window, a specific timestamp, etc).
• A type of object (e.g. user, window, timestamp) is called a class.
• Objects contain data about that object ("attributes") and the class provides functions to manipulate the objects ("methods").
• Methods can provide access to the attributes of an object (e.g. my $title = $window->get_title() and $window->set_title($title);)
• Methods can provide cause actions to be performed. (e.g. $window->close();)