From owner-freebsd-current Sun Mar 24 14:58:19 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id OAA29716 for current-outgoing; Sun, 24 Mar 1996 14:58:19 -0800 (PST) Received: from originat.demon.co.uk (originat.demon.co.uk [158.152.220.9]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id OAA29683 for ; Sun, 24 Mar 1996 14:57:11 -0800 (PST) Received: (from paul@localhost) by originat.demon.co.uk (8.7.5/8.6.9) id WAA02597; Sun, 24 Mar 1996 22:56:30 GMT From: Paul Richards Message-Id: <199603242256.WAA02597@originat.demon.co.uk> Subject: Re: fseek To: joerg_wunsch@uriah.heep.sax.de Date: Sun, 24 Mar 1996 22:56:30 +0000 (GMT) Cc: freebsd-current@FreeBSD.ORG In-Reply-To: <199603241805.TAA04491@uriah.heep.sax.de> from "J Wunsch" at Mar 24, 96 07:05:26 pm Reply-to: paul@netcraft.co.uk X-Mailer: ELM [version 2.4 PL23] MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk In reply to J Wunsch who said > > Our definitions are ``more correct'', but violate ANSI. They have to > be changed, or we cannot claim them ANSI compliant. (I've also > stumpled across this when importing lint.) I've changed this to be ANSI compliant now. > Btw., since you happen to continue the work on lint... Our libc > includes two different ``struct pmap''s. One is the well-known one > for the portmapper, the other one sneaked in from the VM code and > describes a ``page map'' (or something like this). We need to change > one of them, most likely the latter one, in order to ever get libc > linted. I've once tried to add a ``pmap_t'' typedef for the latter > type, but broke everything with it, since i've apparently overlooked > something. This was about the point where i've stopped attempting to > get lint fully included. I'll make a note of this. > > More nits back from the old days (might or might not still apply): > > has definitions of the form: > > #define ntohl __byte_swap_long Umm, the htonl macro (and I suspect this one) completely and utterly breaks lint. It flags a syntax error for the macro and then aborts some time later because it's unable to recover from it. As an example in our tree look at netstat. I first spotted it lint'ing Apache. originat:usr.bin/netstat% lint * if.c: if.c:168: syntax error if.c:168: __X undefined if.c:168: syntax error if.c:174: case not in switch if.c:182: syntax error if.c:182: __X undefined if.c:182: syntax error if.c:183: netnum undefined if.c:186: sipx undefined if.c:206: case not in switch if.c:215: default outside switch if.c:232: syntax error if.c:232: cannot recover from previous errors It's surprising how much of a mess our code is in actually when you run lint across it. I've only got access to FreeBSD at weekends since I work away during the week but I'll gradually work my way through all this if no-one works on it in the meantime. -- Paul Richards, Originative Solutions Ltd. Internet: paul@netcraft.co.uk, http://www.netcraft.co.uk Phone: 0370 462071 (Mobile), +44 1225 447500 (work)