Using system sort order

[sciurius]

Dear monks,
I want my perl program to sort just like the system it is running on.
This is trivial on Linux, but somehow seems complicated on Windows and macOS. I use the following test program

#! perl

use strict;
use warnings;
use utf8;
binmode STDOUT => ':utf8';

my @words = ( "Maria Lenore", "Mária Helena", "María Dolores" );

my @sorted = sort @words;

if ( "@sorted" eq "María Dolores Mária Helena Maria Lenore" ) {
    print("Sorted OK\n");
}
elsif ( "@sorted" eq "Maria Lenore María Dolores Mária Helena" ) {
    print("Sorted C (ASCII)\n");
}
else {
    print("Sort failed: @sorted\n");
}

On Windows, I get

PS C:\Users\Johan> perl locale.pl
Sorted C (ASCII)
PS C:\Users\Johan> perl -Mlocale locale.pl
Sorted C (ASCII)
PS C:\Users\Johan>

Same on macOS.

When I create files with names "Maria Lenore", "Mária Helena" and "María Dolores" they sort correctly in the Explorer (Windows) and Finder (macOS).
How can I have perl use the same sort order as the system?

Windows 10 with Strawberry Perl 5.30
MacOS 10.15 with Perl 5.34.
These versions are a bit old, but I can not imagine I'm the first to try such a sort in 25+ years.

[Arunbear]

On Linux, without the -Mlocale switch, I get the same behaviour that you get on Windows. To get the result you wish, I used Unicode::Collate e.g.

#! perl

use strict;
use warnings;
use utf8;
use Unicode::Collate;

binmode STDOUT => ':utf8';

my @words = ( "Maria Lenore", "Mária Helena", "María Dolores" );
my $col = Unicode::Collate->new(level => 1);

my @sorted = $col->sort(@words);

if ( "@sorted" eq "María Dolores Mária Helena Maria Lenore" ) {
    print("Sorted OK\n");
}
elsif ( "@sorted" eq "Maria Lenore María Dolores Mária Helena" ) {
    print("Sorted C (ASCII)\n");
}
else {
    print("Sort failed: @sorted\n");
}

I had these bookmarked ages ago, but didn't look at them until today:
• Case- and Accent-insensitive Sorting
• What's Wrong with sort and How to Fix It

[sciurius]

Thanks for the suggestion! This does, indeed, give the desired results on both Windows and macOS.

But it does not use the system settings...
See e.g. https://youtu.be/j64acCeati4?t=548 .

For the time being it is already a substantial improvement.

[1nickt]

I can confirm that the sort order on my Mac is what you have posted, but I'm surprised and I feel it's wrong. There's no way I can imagine that á should come before a or í before i with the LC_COLLATE value set to en_US.UTF-8. I think Perl is getting it right here.

[ysth]

It isn't doing anything fancy like sorting all a variants together but sorting the within a by diacritic, it is treating the a's as equal and sorting these strings by the different second words.