There is a standard for Unicode collation. http://www.unicode.org/reports/tr10/

For what you want, perhaps 'LC_COLLATE=C; ls -1 | sort --ignore-case' would suit.

1 members found this post helpful.

allend: I like your suggestion --- very specific rather a vague suggestion. It is much closer to what I dreamt of.

"a" and "A" mix freely, since the cases have been ignored. How nice is an X where LC_COLLATE=X that will do exact what I want. I will check briefly
the site you suggested.

Thanks a lot!

On second thought, sort --ignore-case is really probably want I need. The idea is very simple, I have almost never mix
(>99%?) upper and lower cases. For example, I will have Project-A, Project-B

or project-A, project-B, but

not

project-A, project-b,

or

Project-A, project-B.



The only exception I can think of is

one consistent set of

Project-specialAA, Project-specialBB, Project-specialCC,

and mix with

another consistent set of

Project-SpecialABC, Project-SpecialEFG, Project-SpecialHIJ,

Well, if this rare case happens, I just need to rename the files OR to use two directories.
Perfect yet easy solution is hard to come by.

You need to take the semicolon out. With the semicolon you have two commands, thus requiring the export to get the LC_COLLATE to be passed to the ls command.

"LC_COLLATE=C ls -1" will export LC_COLLATE=C to only one process - the ls.