Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!haven!adm!smoke!gwyn
From: gwyn@smoke.BRL.MIL (Doug Gwyn )
Newsgroups: comp.lang.c
Subject: Re: fwrite man page "error"
Message-ID: <9813@smoke.BRL.MIL>
Date: 8 Mar 89 02:28:50 GMT
References: <939@unh.UUCP> <1463@mcgill-vision.UUCP>
Reply-To: gwyn@brl.arpa (Doug Gwyn (VLD/VMB) <gwyn>)
Organization: Ballistic Research Lab (BRL), APG, MD.
Lines: 18

In article <1463@mcgill-vision.UUCP> mouse@mcgill-vision.UUCP (der Mouse) writes:
>In article <939@unh.UUCP>, jeff@unh.UUCP (Jeffrey E. F. Friedl) writes:
>> I suppose I should tell him that sizeof(char) is !always 1, though.
>	When sizeof is applied to a char, the result is 1; [...].
>I think the first edition contained something similar, though I'm not
>sure.  Do you know of any compilers which disagree?

K&R 1st Ed. Appendix A, generally taken as the de facto pre-ANSI C
standard, says that the units of sizeof are bytes, undefined by the
C language.  It also notes that in all existing (meaning AT&T) C
implementations byte and char are essentially synonymous.

The proposed ANSI C standard officially requires the identity between
the byte (sizeof unit) and the space used to hold a char.  There is
apparently a LOT of existing code that relied on that non-guaranteed
property, so X3J11 decided it would be a useful service to guarantee
it.  (I argued against that, for reasons posted to this newsgroup
some time ago.)