This is slightly off-topic but I can't resist expressing my dislike for this post to which I'm replying. Writing a bogus routine that seems to do what's expected, even with test cases, and without strong indication that it's just a hoax as first reply to a serious question is just despicable. It contributes nothing but confusion. Sure, cute code can be fun but it should be in the right place. The first reply to a serious question is definately not the right place. Making it look like a serious reply doesn't make it any better. If one truly does want to contribute one would also include an explanation of how the cute code works. Otherwise it's just showoff.

The smilie and the "sometimes" hints you about it being bogus, but the "sometimes" can easily be interpreted as "it tries to make a qualified guess but sometimes fails" (in contrast of long, slow but working algorithms), especially since it provides several test cases. I can't blame any new Perl programmer that think "hey, cool, yet another guru trick" when looking at the above routine with all its, to the new Perler, new elements: %_, $%, and ~~. (Compare with the fairly common my $file = do { local (@ARGV, $/) = $filename; <> };, except that works.)

To broaden this post I'd like to encourage everyone to follow Ovid's (I believe, can't find the reference now) example of waiting a while for the question to be answered before posting cute solutions or getting into side-track discussions.

What scares me though is that the post has a fairly high reputation in comparison to the replies which provide solutions that actually work.

Just my thoughts,
ihb