Path: utzoo!utgpu!water!watmath!clyde!ima!haddock!karl
From: karl@haddock.ISC.COM (Karl Heuer)
Newsgroups: comp.lang.c
Subject: Re: Character handling functions -- Jan 88, dpANS
Keywords: isalpha islower isupper locale confusion
Message-ID: <3223@haddock.ISC.COM>
Date: 29 Mar 88 22:11:32 GMT
References: <6192@dhw68k.cts.com> <7571@brl-smoke.ARPA>
Reply-To: karl@haddock.ima.isc.com (Karl Heuer)
Organization: Interactive Systems, Boston
Lines: 18

In article <7571@brl-smoke.ARPA> gwyn@brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>) writes:
>In article <6192@dhw68k.cts.com> david@dhw68k.cts.com (David H. Wolfskill)
>>[voices his concerns about isalpha() != (islower() || isupper())]
>
>Yes, this is right.  Not all character sets have a meaningful concept
>of "case".  (Consider Chinese.)  Instead of arbitrarily picking either
>lower or upper (or both), the implementor can simply tell the truth.

I agree that it's good to allow for such neither-case letters, but both-case?
Is it really intentional to allow isupper() and islower() to be simultaneously
true?

On a related note, I can see where it might be useful for me to define a
locale which reflects the native alphabet only, specifically excluding the
C-locale alphabet.  (E.g. I might want isalpha('w') to be false.)  The dpANS
doesn't allow this.  Why not?

Karl W. Z. Heuer (ima!haddock!karl or karl@haddock.isc.com), The Walking Lint