use locale; on ActiveState WIN32

heth has asked for the wisdom of the Perl Monks concerning the following question:

Hi,

I am a teacher on a Danish Technical High School, and had a Perl Network Management Course this week. On of the students submitted a question that we solved, but I could not explain it.

My student was sorting words, but sort() got it all wrong. I discovered that he used use locale; which I know can give problems. We solved his problem, but I could not explain what happened.

Environment: ActiveState Perl 5.8.8 build 822 (MSWin32-x86-multi-thread) on Danish XP-pro. The following sample code fails on Windows but not on FreeBSD, Solaris or RedHat. Also tried ActiveState 5.10.0 on XP, which failed as well.

use strict;
use warnings;

my @a = qw(a b c d e f g h i j k l A B C D E F G H I J K L);
my @b = qw(aaa aab aba abb Laa lab);

no locale; print "no locale : ";
print sort(@a),"\n";
use POSIX; print "POSIX     : ";
print sort(@a),"\n";
use locale;print "use locale: ";
print sort(@a),"\n";

no locale; print "no locale : ";
foreach (sort(@b)){printf "%-4s", $_;}
print "\n";
use POSIX; print "POSIX     : ";
foreach (sort(@b)){printf "%-4s", $_;}
print "\n";
use locale;print "use locale: ";
foreach (sort(@b)){printf "%-4s", $_;}
[download]

This code produces the following output on XP:

no locale : ABCDEFGHIJKLabcdefghijkl
POSIX     : ABCDEFGHIJKLabcdefghijkl
use locale: aAbBcCdDeEfFgGhHiIjJkKlL
no locale : Laa aaa aab aba abb lab 
POSIX     : Laa aaa aab aba abb lab 
use locale: aba abb lab Laa aaa aab
[download]

Does anybody know why the sort with use locale produces aba abb lab Laa aaa aab.

Best regards

Henrik Thomsen

Comment on use locale; on ActiveState WIN32 Select or Download Code

Replies are listed 'Best First'.
Re: use locale; on ActiveState WIN32 by Anonymous Monk on Oct 11, 2008 at 18:31 UTC
danish alphabet :)? Try this snippet from perllocale print "\n-------------------\n"; use locale; print +(sort grep /\w/, map { chr } 0..255), "\n"; print "\n-------------------\n"; no locale; print +(sort grep /\w/, map { chr } 0..255), "\n"; print "\n-------------------\n"; __END__ no locale : ABCDEFGHIJKLabcdefghijkl POSIX : ABCDEFGHIJKLabcdefghijkl use locale: aAbBcCdDeEfFgGhHiIjJkKlL no locale : Laa aaa aab aba abb lab POSIX : Laa aaa aab aba abb lab use locale: aaa aab aba abb Laa lab ------------------- _µ01¹2²3³456789aAªáÁàÀâÂäÄãÃåÅæÆbBcCçÇdDðÐeEéÉèÈêÊëËfFƒgGhHiIíÍìÌîÎïÏj +JkKlLmMnNñÑoOºóÓòÒôÔöÖõÕøØœŒpPqQrRsSšŠßtTþÞuUúÚùÙûÛüÜvVwWxXyYýÝÿŸzZžŽ ------------------- 0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz ------------------- [download] It has something to do with LC_COLLATE :)	[reply] [d/l]
Re^2: use locale; on ActiveState WIN32 by heth (Sexton) on Oct 11, 2008 at 19:45 UTC
hi :-) Output: `------------------- _µ01¹2²3³456789aAªáÁàÀâÂãÃbBcCçÇdDðÐeEéÉèÈêÊëËfFƒgGhHiIíÍìÌîÎïÏjJkKlLm +MnNñÑoOºóÓòÒôÔõÕœŒpPqQrRsSšŠßtTþÞuUúÚùÙûÛvVwWxXyYýÝÿŸüÜzZžŽæÆäÄøØöÖåÅ ------------------- 0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz -------------------` [download]	[reply] [d/l]
Re: use locale; on ActiveState WIN32 by syphilis (Archbishop) on Oct 11, 2008 at 23:17 UTC
Heh ... I get the same output as you for @a, but for @b with `use locale` I get something different: `use locale: aaa aab aba abb Laa lab` [download] Based on the @a output I would expect: `use locale: aaa aab aba abb lab Laa` [download] which is different to what both you and I are getting. Someone here will probably understand what's happening ... not me, but. Cheers, Rob	[reply] [d/l] [select]
Re: use locale; on ActiveState WIN32 by Narveson (Chaplain) on Oct 12, 2008 at 10:01 UTC
The sequence 'aa' is being sorted as if you meant å (U+00E5, LATIN SMALL LETTER A WITH RING ABOVE). Am I right in thinking this letter comes quite a bit later than 'a' in Danish dictionaries?	[reply]
Re^2: use locale; on ActiveState WIN32 by heth (Sexton) on Oct 12, 2008 at 14:01 UTC
Hi Narveson Thank you, I've solved the challange. Great that a American should solve a problem with my native alphabet. Thank you, and have a nice day :-) :-) Henrik	[reply]