sorting fails on Japanese Unicode characters

[Rolf Leggewie]

there seems to be some oddity in interpreting some Japanese unicode characters in bash and how they should be sorted.

$ ls -1 /tmp/*.txt
/tmp/⑥-test.txt
/tmp/⑤-test.txt
/tmp/④-test.txt
/tmp/①-test.txt
/tmp/③-test.txt
/tmp/②-test.txt
$ ls -1 /tmp/*.txt|sort
/tmp/⑥-test.txt
/tmp/⑤-test.txt
/tmp/④-test.txt
/tmp/①-test.txt
/tmp/③-test.txt
/tmp/②-test.txt

This is while booted into an uptodate precise system.

[Rolf Leggewie (r0lf)]

assigning to bash and coreutils packages for now for confirmation and triage

[Fumihito YOSHIDA (hito)]

'sort' command does not support "human graspable sorting" in unicode environments.
http://www.gnu.org/software/coreutils/faq/#Sort-does-not-sort-in-normal-order_0021

I suggest solution, sort with "LC_ALL=C" variables (or set in your aliases).

This is bad spec, but this is unalterable by historical reason.

$ ls -1 | sort
①
③
②
②-test.txt
⑤-test.txt
④-test.txt
⑥-test.txt
①-test.txt
③-test.txt

$ ls -1| LC_ALL=C sort
①
①-test.txt
②
②-test.txt
③
③-test.txt
④-test.txt
⑤-test.txt
⑥-test.txt