http://www.perlmonks.org?node_id=866794


in reply to Re^7: perllocale weirdness, bug, or...?
in thread perllocale weirdness, bug, or...?

Thanks for the tip - I really hoped it might solve the issue, but unfortunately it does not.

For the file test.txt (TAB separated fields) being
a_2 2 a2 1
I still get
$ sort -k 1,1 test.txt | cut -f 1 a2 a_2
whereas when there is no second field then the result is reversed,
a_2 a2
Looks like the problem 'survives'.
---

Replies are listed 'Best First'.
Re^9: perllocale weirdness, bug, or...?
by choroba (Cardinal) on Oct 22, 2010 at 12:33 UTC
    Looks like two level sort: on the first level, it ignores the _, but if the strings are different only thanks to the underscore, it is used in the second level.
    $ echo $'a_2\t2\na2\t1' | sort -k1 a2 1 a_2 2 $ echo $'a_2\t2\na2\t1' | sort -k1,1 a2 1 a_2 2 $ echo $'a2\t2\na_2\t1' | sort -k1,1 a2 2 a_2 1 $ echo $'a2\t2\na_2\t1' | sort -k1 a_2 1 a2 2