Unicode strange behaviour as a key of hash

[leszekdubiel]

Could anyone tell why the program displays "error 2"? This shouldn't display any error... nor first nor second.

#!/usr/bin/perl -CSDA

use utf8; 
use strict; 
use warnings; 

my %h; 
$h{'Góry'} = 1; 
'Góry' =~ /\A[[:alnum:]]*\z/ or die "error 1"; 
for (keys %h) { 
    $_ =~ /\A[[:alnum:]]*\z/ or die "error 2"; 
}

Leszek Dubiel

[dave_the_m]

It appears to be a bug the optimisation of constant hash keys. The constant key string in $h{'Góry'} is being downgraded from utf8, whereas if you write my $g = 'Góry'; $h{$g} = 1; it works ok. Once the key is downgraded, you then run into the Unicode Bug. This can bypassed in pattern matches using the //u match modifier.

Dave.