Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!lll-crg!lll-lcc!pyramid!decwrl!sun!guy
From: guy@sun.uucp (Guy Harris)
Newsgroups: net.bugs.4bsd
Subject: <sys/types.h> misdeclares "time_t" and "off_t"
Message-ID: <5682@sun.uucp>
Date: Thu, 31-Jul-86 01:01:59 EDT
Article-I.D.: sun.5682
Posted: Thu Jul 31 01:01:59 1986
Date-Received: Fri, 1-Aug-86 01:10:45 EDT
Distribution: net
Organization: Sun Microsystems, Inc.
Lines: 30

Index:	sys/h/types.h 4.3BSD

Description:
	In every UNIX system except 4BSD (and maybe 32V), "time_t" and
	"off_t" are declared as "long".  In 4BSD, they are declared
	as "int".  This doesn't affect code on "typical" systems, but
	it does give "lint" gastric distress.  The manual page for
	"time", for instance, says that it takes a pointer to a "long"
	as an argument.  However, the "lint" library says it takes a
	pointer to a "time_t".  This means that code written to conform
	to the specifications of the 4BSD (and V7, and S3, and S5, and...)
	manual doesn't pass "lint".

	Furthermore, this means that programs that use "int" values where
	a "time_t" or "off_t" is expected will pass "lint" on a 4BSD
	system; they won't pass "lint" on other systems, and if "int"s
	aren't the same size as "long"s they won't *work* there either.
Repeat-By:
	Compare <sys/types.h> with any other UNIX variant's version.
Fix:
	Change them to be "typedef"fed to "long" rather than "int".
	This will cause some code in the kernel - and probably some
	user code as well - not to "lint", but that can be fixed, and
	should be, since the existing code is incorrect.  (No, this
	fix won't cause any incurable "lint" problems, I successfully
	"lint"ed a version of our kernel with the correct "typedefs".)
-- 
	Guy Harris
	{ihnp4, decvax, seismo, decwrl, ...}!sun!guy
	guy@sun.com (or guy@sun.arpa)