the reference to rm -rf's at the bottom of an obfuscation made me think twice before running it

If an obfuscator wanted to zap you like that, he'd probably obfuscate the rm -rf somehow. You'd have no warning unless you de-obfuscated the code before running it, or read comments by others who either de-obfuscated it or got bitten. There are an infinite number of ways to obfuscate code, as I'm sure you're aware if you follow this section closely.

My suggestion is to keep an unprivileged account around for running untrusted code. I'm generally not a big advocate of unprivileged accounts; for normal, everyday use I feel that they cause more inconvenience than they're worth[1]. However, for running untrusted code, or code that processes untrusted data from the internet (especially, any kind of server code), an unprivileged account can save you a lot of grief. If you don't trust an obfu (or whatever other code you don't trust) run it as a user with no privileges, no access to your home directory with your data.

That said, I'll admit that with obfuscations on Perlmonks I often don't bother, especially if there are already positive comments by monks whose names I recognize. I haven't been bitten yet...

[1] I say this not to persuade anyone that it's true, nor to start an argument about it (I'm tired of that argument, believe me), but to point out that even someone who holds this view, such as myself, still sees the value of an unprivileged account for running untrusted code. Where you draw the line in terms of what software you choose to trust is another matter.

$;=sub{$/};@;=map{my($a,$b)=($_,$;);$;=sub{$a.$b->()}}
split//,".rekcah lreP rehtona tsuJ";$\=$ ;->();print$/
[download]