Perl - Sorting In UNIX Different From Windows

Question

I'm having issues sorting in Perl having different results in Windows and Unix.

The characters are: a - _ 1 2

In Windows: _ 1 2 - a
In Unix: _ - 1 2 a

I'm guessing the locale has something to do with this - what can I do to make the Unix sort match the Windows sort?

Thanks!

score 2 · Accepted Answer · answered Feb 16 '12 at 03:21

2

The docs say:

*** WARNING *** The locale specified by the environment affects sort order. Set LC_ALL=C to get the traditional sort order that uses native byte values.

so use

LC_ALL=C sort ...

Example:

$ perl -E'say for @ARGV' a - _ 1 2 | LC_ALL=en_US.UTF-8 sort
_
-
1
2
a

$ perl -E'say for @ARGV' a - _ 1 2 | LC_ALL=C sort
-
1
2
_
a

answered Feb 16 '12 at 03:21

ikegami

367,544
15
269
518

These are still not the same as the results in Windows. I need it to match Windows' sort - is there a collation on Unix that will do this? – Michael Sweetser Feb 16 '12 at 16:58
@Mike, oo, I see that Windows's uses a local too. Well, you'll have to find out what local your Windows uses and see if there's an equivalent one on your unix system, and use it instead of `C` as I showed. I wish you luck! – ikegami Feb 16 '12 at 18:32
That seems to be where I'm at - English_United States.1252 seems to be the Windows one, but I can't yet find a Unix counterpart. – Michael Sweetser Feb 16 '12 at 19:29
Originally, the following was a comment, but, for more exposure, I just made a new question out of my comment. I then erased the comment, because I don't want to have people answer it twice. I originally addressed comment to @tchrist, because he strikes me as a guru on the matter. http://stackoverflow.com/questions/15013515/multilingual-text-sorting-in-perl-on-windows-using-locale – egilchri Feb 21 '13 at 22:34

choroba · Answer 2 · 2012-02-17T08:17:51.840

0

If you do not want to use locale, comment out the line containing

use locale;

Without such a line, sort in Perl should behave the same on both Windows and Unix.

You can also add

no locale;

before the sort (or, better, enclose the sort into a block starting with it).

edited Feb 17 '12 at 08:17

answered Feb 16 '12 at 10:06

choroba

231,213
25
204
289

I don't have "use locale" in the scripts at all. – Michael Sweetser Feb 16 '12 at 16:58
@Mike all it takes is one module using it to change it everywhere. – Brad Gilbert Feb 16 '12 at 21:57
My script is not using any modules at all. I tried no locale anyway with no changes. – Michael Sweetser Feb 17 '12 at 20:24
@Mike: I'm getting the same results for `perl -le 'print sort qw/a - _ 1 2/'`. – choroba Feb 18 '12 at 03:11

Perl - Sorting In UNIX Different From Windows

2 Answers2