On Thu, 03 Apr 2008, David Holland wrote:
> I think another valid question to raise is whether this global
> variable needs to continue to be supported...
It's required by POSIX
but I don't see it documented in NetBSD's tzset(3) man page or anywhere
else obvious. We do have timezone(3), but that's a function, not a
variable, and it's not in POSIX.
The loop starting at line 281 of src/lib/libc/time/localtime.c looks
suspicious. I don't immediately see what sp->typecnt really means, but
it does appear that that timezone and altzone may be set more than once.
--apb (Alan Barrett)