As an aside, you might not want to commit the: my %uniq; @uniq{@arr} = (1) x @arr; @arr = keys %uniq; idiom to memory too closely, because it's broken :)

my @arr = ("what", "the", "the", {"heck" => "I"},
           {"heck" => "I"}, "don't", "get", "this");

my %uniq; @uniq{@arr} = (1) x @arr; @arr = keys %uniq;
print join "\n", @arr;
[download]

prints:

this
don't
the
HASH(0x1002f4e0)
get
HASH(0x1002116c)
what

perldoc -q duplicate offers a few ways to do this depending on what you mean by "removing duplicates", but upon brief inspection it doesn't look like any of those would catch this problem either.

Hmmm...an interesting golf perhaps? :)

(Here's what it looks like in Python, if that helps any)