Path: utzoo!attcan!uunet!usenix!std-unix
From: karl@IMA.IMA.ISC.COM (Karl Heuer)
Newsgroups: comp.std.unix
Subject: Re: _POSIX_VDISABLE
Message-ID: <399@usenix.ORG>
Date: 24 Jul 90 22:03:57 GMT
References: <10462@cs.utexas.edu>
Sender: std-unix@usenix.ORG
Reply-To: karl@IMA.IMA.ISC.COM (Karl Heuer)
Organization: Interactive Systems, Cambridge, MA 02138-5302
Lines: 16
Approved: jsq@usenix (Moderator, John Quarterman)

In article <10462@cs.utexas.edu>:
>From: mcgrath%tully.Berkeley.EDU@ucbvax.Berkeley.EDU (Roland McGrath)
>The problem is that the wording of the standard and the sysconf, pathconf, and
>fpathconf functions were designed for boolean options and for limits which are
>required to be positive integers.  In these cases, -1 is a reasonable
>out-of-range value.  But _POSIX_VDISABLE is a character value, not restricted
>to any specific range, and doesn't fit in right.

The documented use of _POSIX_VDISABLE is to store it in an object of type
cc_t.  This is defined in 7.1.2.1 to be an unsigned integral type; hence -1
is indeed out-of-band.  Unless you're trying to make cc_t an unsigned long and
have _POSIX_VDISABLE be ULONG_MAX, you should be okay.

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

Volume-Number: Volume 20, Number 143