>  Currently I have a hash, which key is the fully qualified
>  pathname/file and its value is the md5sum (by the way: Is there a
>  md5sum-generator inside Ruby ?) of that file.

Key = md5sum, values = array of filenames
(feed with "hash[md5(filecontents)] << filename")
this way you know the duplicates names and can find if they are duplicated by 
doing something like

hash.each { |x| puts "dups" if x.size > 1 }

It may be not quite as fast esp. if there are no duplicates, but should work.

-- 
keep in touch. berkus. -- http://lye.upnet.ru/