The meaning of [A-Z], on a gnu system, depends on the locale used. The letters between [A-Z] depend on the sorting order defined in the locale, and I guess that's defined by language in unicode somehow; but I don't know how much there is a decision by Unicode on that. For example, in Catalan we have that [A-Z] array defined as [aAbBcCdDeE...]. And imagine: [A-Z] doesn't include [a]. :)
Using "LANG=C" before evaluating any [A-Z] expression should give you the usual C meaning of only capitals. 2008/2/9, Charles Forsyth <[EMAIL PROTECTED]>: > this subject line reminds me: > > i'm fed up on Linux (Ubuntu) with bash messing up both file name matching > and 9term editing and i'd like to switch to p9p's rc. > (which twit decided that [A-Z] could ever be case-insensitive with a > case-sensitive file store?) > is it just a matter of chsh (after changing /etc/shells)? > given my experience with Linux, that seems implausibly straightforward. > i tried googling for some a little while ago but without success. > >
