kj wrote: > sort behaves erratically with underscores: > > % ( echo _c; echo __; echo _a ) | sort > __ > _a > _c
Sort uses your current locale setting (e.g. LANG) to determine the character collation sequence. You probably have LANG set to a dictionary sort order. In dictionary sort order case is folded and punctuation is ignored. See this FAQ entry for more information: http://www.gnu.org/software/coreutils/faq/#Sort-does-not-sort-in-normal-order_0021 > How can I get sort to treat _ consistently? (I don't have a strong > preference for either _ < a or a < _ as long as it is consistent.) You can solve this by setting a standard sort order instead of a non-standard dictionary sort ordering locale. "C" (or the "POSIX" alias) is the normal standard one. LANG=C sort Personally I set the following in my own environment to get UTF-8 but force a standard sort ordering regardless. export LANG=en_US.UTF-8 export LC_COLLATE=C Bob
