First thing, the way you have it set up to check timestamps,
the line if ($time_to_die >= ($current_time - $last_time)) {
seems to suggest that $time_to_die is in seconds, as both
$current_time and $last_time are in seconds, but I believe
you were expecting $time_to_die in minutes, so you may
want to throw a *= 60 in there or something.
One thing I was considering about this is that there isn't
really a need to throw the timestamp into a file, since
basically you're just storing the exact same number as is
already contained in the file's modification time. Since
you're already effectively performing a stat via the -e
test, you can then just use the cached filehandle _ to
obtain the modification time and cut out an additional
file opening/reading/writing.
use Benchmark;
use strict;
my $time_to_die = 15;
my $time_file_1 = "time1.txt";
my $time_file_2 = "time2.txt";
my $current_time = time(); # kept this outside the benchmark
timethese(100000, {
'Time In File' => \&timeinfile,
'File Mod Time' => \&filemodtime
});
sub timeinfile {
if (-e $time_file_1) {
open (TIME, "$time_file_1") || warn "Can't open Time file: $!";
my $last_time = <TIME> || warn "Unable to get timestamp from fil
+e: $!";
close TIME;
if ($time_to_die >= ($current_time - $last_time)) {
open (TIME, ">$time_file_1");
print TIME $current_time || warn "Unable to write timestamp t
+o file: $!";
close TIME;
}
else {
unlink($time_file_1) or warn "Can't unlink file: $!";
}
}
else {
open (TIME, ">$time_file_1");
print TIME $current_time;
close TIME;
}
}
sub filemodtime {
if(-e $time_file_2) {
my $last_time = (stat(_))[9];
if($time_to_die >= ($current_time - $last_time)) {
open (TIME, ">$time_file_2");
close TIME;
}
else {
unlink($time_file_2) or warn "Can't unlink file: $!";
}
}
else {
open (TIME, ">$time_file_2");
close TIME;
}
}
Benchmark: timing 100000 iterations of File Mod Time, Time In File...
File Mod Time: 11 wallclock secs ( 6.97 usr + 3.21 sys = 10.18 CPU)
Time In File: 48 wallclock secs (21.26 usr + 25.77 sys = 47.03 CPU)