Re: [Nmh-workers] Compiler warnings and signed vs. unsigned char, again

[Ken Hornstein]

>No, I mean that gcc is highlighting a bit of lint and pointing out that
>passing a char to ctype is non-portable and may bite elsewhere, even if
>the current platform's native char is unsigned, e.g. ARM.  Altering char
>to be unsigned doesn't stop it trying to be helpful AIUI.

Well, AFAICT gcc doesn't complain about the issue with ctype at
all.  (Does it?  Let me check.  No, it does not).  What really got
this started was the buildbot on Solaris complaining about a ton of
signed-versus-unsigned char issues.  That let me to research how we
could get rid of those ... and here we are :-)

Well, okay, let me rephrase that ... it doesn't on my system.  On
NetBSD it does because the ctype macros are (in many cases) indexes
into an array.  But as I understand it altering char to be unsigned
would actually solve that particular warning as well.

--Ken