Red Hat Bugzilla – Bug 10294
folded collating sequence for en_US
Last modified: 2008-05-01 11:37:55 EDT
At least on Alpha, but maybe also on other platoform a collating sequence supplied in /usr/share/locale/en_US/LC_COLLATE seems to fold upper and lower case thus rendering sorting utilities rather ineffective. The reason I suspect that is that in order to restore some sanity in sort one has to unset both LANG and LC_ALL environment variables. If at least one of these is set to en_US then sort colapses. Curiously enough 'ls' also appears to consult LC_COLLATE if one of variables above is set but it is not affected by this bogosity. Nichal Jaegermann michal@harddata.com
(changed to all platforms, it affects all of them)
assign to jakub
That's how en_US collation should look like (the case of characters matters only when the strings are equal in case insensitive comparison). Basically, if you want ASCII sorting, use POSIX/C locale, if you want the sorting people have been doing for centuries, use your locale. ls is still using strcmp while it should use strcoll, it will be changed in the future.