https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=284864
--- Comment #12 from w...@psr.com --- (In reply to Andriy Gapon from comment #11) > Inclined to close this as 'not a bug'. I agree it's not a bug. However ... > This is not about case sensitivity and, thus, -f flag really. It kind of is. > The manual page already has several references to collation order and > LC_COLLATE. Indeed it does, such as "... are performed lexicographically, according to the current locale's collating rules" said right there in the first paragraph. I had no idea that could mean "including upper/lower case differences not mattering if the remainder of the sort field differs (for locales other than C)". > Maybe they should be more prominent. I don't think that would help. I now understand the collating rules better, you had to rediscover them, they do indeed depend on the locale, etc. Someone who doesn't already know all that isn't going to realize that locale rules (except for C) usually result in case-independent sorting. Unless that's said explicitly, merely saying "it depends" more prominently won't make that apparent. I think the current description of -f suggests that one uses the switch to control case-independent sorting and that, absent -f, one can expect more C-like case-dependent sorting. Except for C locale and sort fields that are identical except for case (level 3), that conclusion is incorrect. Thus I think it merits some kind of comment under -f not to expect that (except for C) OR an explicit comment somewhere that, except for C locale, upper/lower case is less important to sort order than what follows on the line and thus may match case-independently. Just my opinion. I had no idea locale collation rules do what they do, and the only reference to case-(in)dependent sorting was the description under -f, and I came to the wrong conclusion. -- You are receiving this mail because: You are the assignee for the bug.
