From owner-cvs-all Fri May 18 8:35:31 2001 Delivered-To: cvs-all@freebsd.org Received: from khavrinen.lcs.mit.edu (khavrinen.lcs.mit.edu [18.24.4.193]) by hub.freebsd.org (Postfix) with ESMTP id DCD5C37B42C; Fri, 18 May 2001 08:35:19 -0700 (PDT) (envelope-from wollman@khavrinen.lcs.mit.edu) Received: (from wollman@localhost) by khavrinen.lcs.mit.edu (8.9.3/8.9.3) id LAA73284; Fri, 18 May 2001 11:35:02 -0400 (EDT) (envelope-from wollman) Date: Fri, 18 May 2001 11:35:02 -0400 (EDT) From: Garrett Wollman Message-Id: <200105181535.LAA73284@khavrinen.lcs.mit.edu> To: Bruce Evans Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/sys/arm/include ansi.h src/sys/i386/include ansi.h src/sys/powerpc/include ansi.h In-Reply-To: References: <20010517225829.B68924@dragon.nuxi.com> Sender: owner-cvs-all@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG < said: > According to wollman, POSIX (.1g) will require at least 8-bit chars > (char == uint8_t). Of course, the compiler can always fake it except > for accesses to magic (device) memory. It's actually a bit more complicated than that. The networking specification (.1g) was written to harmonize with C89, and required the presence of a uint8_t which was subtly different from what was eventually standardized in C99. (Specifically, C99's version does not allow padding bits -- i.e., bits in the memory representation which do not participate in arithmetic operations -- whereas 1003.1g's version does.) The Austin Group was faced with a choice: we could invent lots of new types and semantics for the networking functions in order to work around the C99's more restrictive definition, or we could require eight-bit characters. Given the small and rapidly vanishing population of non-eight-bit machines, most of which never ran anything even vaguely like UNIX, we chose the latter option, so the current draft now requires uint8_t, uint16_t, and uint32_t to exist. The principal opposition to this approach came not from old PDP-10 hackers, but from people who wanted *16*-bit characters. -GAWollman To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message