I recently had a big revelation at (jeffa) 3Re: Whether to use local(), hopefully this post will remove the vagueness from your mind as well.

UPDATE:
Oh $#!^ - you are Sam Tregar. Scratch that last sentance! :O

jeffa

L-LL-L--L-LL-L--L-LL-L--
-R--R-RR-R--R-RR-R--R-RR
B--B--B--B--B--B--B--B--
H---H---H---H---H---H---
(the triplet paradiddle with high-hat)

I use it any time I would have used "use vars" in the past - to access package globals. The nice thing is that I can scope where that access is legal. So, if I've got a package global %CACHE and two routines read_cache and write_cache I can:

sub read_cache {
  our %CACHE;
  ...
}

sub write_cache {
  our %CACHE;
  ...
}
[download]

And they'll access the same variable. But if I start refering to %CACHE inside some other routine I get an error just like if %CACHE had been a lexical. I like that.

So, I guess I'm pretty happy with our. The only reason I want more education is that I learned about lexical and package variables as totally separate!!! and now they ain't... Which is cool, but also pretty bizarre.

-sam

I don't like to declare the same variable more than once.

sub read_cache {
  our %CACHE;
  ...
}

sub write_cache {
  our %CACHF;
  ...
}
[download]

Oops. It's so easy to slip... But on the other hand, you have to misspell the variable throughout all of &write_cache, which usually isn't too probable. But for short routines--like get/set routines--it's a not too far-fetched scenario.

For those times I prefer to use

{
  our %CACHE;

  sub read_cache {
    ...
  }

  sub write_cache {
    ...
  }
}
[download]

The only reason I want more education is that I learned about lexical and package variables as totally separate!!! and now they ain't... Which is cool, but also pretty bizarre.

I agree, it does indeed look like globals has become... something else. But you can relax--they haven't. They're still totally separate. What our() does is to declare a lexical alias for the global variable.

Cheers,
-Anomo