Here is a simple program (it puts a value to the database and then gets the value back):

use Encode qw/decode is_utf8/;
use DBI;
my $dbh = DBI->connect("dbi:Pg:dbname=ab", "***", "***", { RaiseError 
+=> 1, AutoCommit => 0 });
my $a = decode('iso8859-1', "\x{92}");
$dbh->do("CREATE TABLE a (a text)");
$dbh->do("INSERT INTO a(a) VALUES (?)",{}, $a);
my($b) = $dbh->selectrow_array("SELECT * FROM a");

if($b eq $a){
    print "Equals a: $a, b: $b\n";
}else{
    print "Not equals a: $a, b: $b\n";
}
print "a is_utf8: " , is_utf8($a), "\n";
print "b is_utf8: " , is_utf8($b), "\n";
$dbh->rollback;
$dbh->disconnect;
[download]

And here it's output:

Not equals a: ’, b: Â’
a is_utf8: 1
b is_utf8:
[download]

Suprise! The value that we get back is not equal to what we fed to the database. The database is PostgreSQL with UTF8 as the main charset. I am not sure if I did something wrong, or what should be the behaviour of this code snippet, but this struck me as very suprising.

Update: Adding _utf8_on($b) sets the UTF8 flag on $b and then indeed $a eq $b.

Update: The perl version is 5.8.3 on Linux (thanks mirod).

Update: After some more meditation on this fact I see that it is not much different from the case when you feed lets say string '5.0' into a numeric field and when you get it back you have '5'. You just need to remember that the UTF8 flag is lost when the value goes into the database.