> where chomp? did the initial setup for the chomp, and returned true if 
> the string would be modified, and then chomp! could use that work to
> change the string.
> On performance tests I've been doing, string object creations are a
> BAD THING, so eliminating that dup would be good.

btw, perl does copy-on-write, which eliminates most of the overhead of
string construction; and you can't modify a string value in Perl, which is
a valuable guarantee, IMHO. in Perl, when you do chomp, it actually does
$_=$_.chomp when the reference count is over 1, but when it's only 1 it
does $_.chomp!, which is faster. This is done behind your back and you
can't control it, which ends up being better for your own health in the
long run. =) This actually happens for most operations on strings.

This is the best justification I'd give in favour of reference counting.
Not that I actually like reference counting.



Mathieu Bouchard