Here's one way. The sequence of strings it produces is non-repeating, and, though very long, not infinite.

{
  my $UNIQ_ID = 'a';
  sub next_id { return $UNIQ_ID++ }
}
[download]

  $str = time();
[download]

But seriously, if need this for any kind of security purpose, then you want something that isn't predictable either. There was a good discussion of this at Randomizing Unique ID?. I'll boil it down for you:

Don't re-seed (as discussed here). The default seed is very good, in versions of Perl since 5.004.

If you do choose to re-seed, incorporate the previous seed:

srand( rand(~0) ^ $mySeed );
[download]

That said, use some combination of time, PID (Process ID, or $$), and rand.

Randomizing Unique ID? also contemplates changing the base of the number (making it a string); this resulted in the Base Conversion Utility.

Unfortunately even this scheme is not perfect. You have a couple issues:

• How many unique strings are you trying to make?
• Are they from the same run of the script, or different runs?
  $$ won't change in one run of a script.
  
• How often is it run?
  time only provides resolution down to the second.
  It also can go backwards.*
  On Win32 you have Win32::GetTickCount(), but that repeats every month or so.
  
• There's a chance, ever so slight, that rand will repeat itself when time and pid have not changed.

So the bottom line is, use as much of the output from rand as you can, to reduce the probability of a repeated string.

* Yes, time can go backwards.
"How?"
Well, time uses the system clock, which has a tendency to drift. There are several mechanisms in use to keep the clock right. Some of these methods attempt to return the clock to some standard (UTC) by manually manipulating the drift (changing the number of ticks/sec to get the clock to speed up or slow down). But some systems actually re-set the clock to UTC. If the clock was running fast this re-set will move time backwards.

If you're wanting this for a temporary filename, consider using POSIX::tmpnam or IO::File::new_tmpfile.

If you want a unique name for some sort of data entity, consider using Symbol::gensym instead.

sub generate_random_string {
    my $res = '';
    $res.= chr int rand(26)+96 for 1..8;
    return $res;
}
[download]

my $h={};
sub generate_new_unique_random_str{
my $result;
while(exists $h->{$result=generate_random_string()}) {};
return $result;
}
[download]

The operative word is "unique" though not "random." So I think one of the UID generating modules, like Data::UUID (and there are two or three others IIRC), would be the right answer.

Originally posted as a Categorized Answer.

Originally posted as a Categorized Answer.

sub generate_random_string {
    my $res = '';
    $res.= chr int rand(26)+96 for 1..8;
    return $res;
}
[download]

my $h={};
sub generate_new_unique_random_str{
my $result;
while(exists $h->{$result=generate_random_string()}) {};
return $result;
}
[download]

Originally posted as a Categorized Answer.

sub generate_random_string {
    my $res = '';
    $res.= chr int rand(26)+96 for 1..8;
    return $res;
}
[download]

my $h={};
sub generate_new_unique_random_str{
my $result;
while(exists $h->{$result=generate_random_string()}) {};
return $result;
}
[download]

Originally posted as a Categorized Answer.

my $word = Data::GUID->new()->as_base64();
[download]