Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!mimsy!chris
From: chris@mimsy.UUCP (Chris Torek)
Newsgroups: comp.unix.xenix
Subject: Re: ENOTTY sometimes spurious
Message-ID: <6309@mimsy.UUCP>
Date: Fri, 17-Apr-87 00:04:19 EST
Article-I.D.: mimsy.6309
Posted: Fri Apr 17 00:04:19 1987
Date-Received: Sun, 19-Apr-87 09:52:35 EST
References: <143@sds.UUCP> <415@psu-cs.UUCP>
Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
Lines: 16

In article <415@psu-cs.UUCP> curnutt@psu-cs.UUCP (Bryan Curnutt) writes:
>We're using SCO Xenix on a Sperry IT, and I've found that fprintf(),
>when used with text files, will always set errno to ENOTTY and return
>a non-zero value.  What makes this more interesting is that the
>fprintf() DOES WORK.  I called SCO about this, and they told me
>to ignore it.

They are right, for this is not a bug.  In System V, fprintf returns
the number of characters written, `or a negative value if an output
error was encountered'.  C library (manual section 3) routines,
unless otherwise specified, do not promise anything about `errno'.
It is not correct to use errno even if fprintf returns an error
indication (although it may be better than nothing).
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7690)
UUCP:	seismo!mimsy!chris	ARPA/CSNet:	chris@mimsy.umd.edu