In a legacy Perl module where I currently fix bugs at work (yes, the same), there is a certain hash that is accessed globally in the main namespace. To protect the innocent, let the name of the hash be foo. (In reality, it did have a descriptive name.) In a certain part of the file, I noticed the following two consecutive lines:
%foo ={};
unlink %foo;
Unless this is very arcane magic, this does nothing what the original programmer thought it would. In scalar context, the above hash evaluates to "1/8" on Perl 5.8.8 in Linux, so this would remove the file 8 in directory 1 in the current working directory. And no, unlink's return value or errors in deleting the "file" are not checked.
I'm guessing the intent was to empty the hash, then to make really, really sure that it is empty by calling some deletion function... What do you make of it? Why would a programmer write something like this? The funniest explanation will win.
UPDATE: Well, you learn something new every day. unlink takes a list as an argument, so the above would actually attempt to delete a file named HASH(0x814dc30) (the numbers may very in your machine). Still, assigning a hash reference into a hash is not what he meant to do either, that's for sure.