perl5.8.0 crazy strings

Can anyone reproduce this? I am on Linux Mandrake 8.2 and perl8.2.0  prints 'no 
match' when the following script (saved as utf-8) is executed. Is it a bug or a 
feature? On perl5.6.1 works as expected.

use utf8;
$a = 'España';
$a =~ s/ñ/ñ/;
print $a =~ m/ñ/ ? 'match' : 'no match', "\n";





Gregor Chrupala