From owner-freebsd-current Thu Oct 10 23:17:24 2002 Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 2EF5A37B408 for ; Thu, 10 Oct 2002 23:17:22 -0700 (PDT) Received: from hun.org (hun.org [216.190.27.122]) by mx1.FreeBSD.org (Postfix) with ESMTP id 8A6FA43EAA for ; Thu, 10 Oct 2002 23:17:21 -0700 (PDT) (envelope-from attila@hun.org) Received: by hun.org (Postfix, from userid 1001) id E6C9C57A80; Fri, 11 Oct 2002 06:17:20 +0000 (GMT) Date: Fri, 11 Oct 2002 06:17:20 +0000 (GMT) Message-Id: <20021011061720.CjK362410@hun.org> From: Daniel Flickinger X-Mailer: AttilaMail with XEmacs & Postfix on FreeBSD 5.0-CURRENT X-Ballistic: N 37.218497 W 113.614979 X-Address: 31 N 700 E, St George UT 84770-3028 X-Squawk: (435) 680-0750 X-No-Archive: yes In-Reply-To: <20021011034541.GA37429@xor.obsecurity.org> References: <20021010230252.Kg3u53327@hun.org> <20021010201920.zsql51239@hun.org> <20021009030149.Juib29905@hun.org> To: Kris Kennaway Cc: FreeBSD-CURRENT Subject: Re: 3+ days: syntax error in makedevs.c (sysinstall) Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1; name="text" Content-Transfer-Encoding: 7bit Content-Disposition: inline Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Sent: Thu, 10 Oct 2002 20:45:41 -0700 by Kris Kennaway = on thu, oct 10, 2002 at 11:02:52pm +0000, daniel flickinger wrote: = = > + the 'installincludes' thing appears to be unnecessary and can possibly = > + cause problems. = > = > well, you can rattle that around with phk, too! [g] = > = > installincludes puts the corresponding headers in = > place for the compiler. theoretically, buildworld = > -i declarations cover the game... but.... = = yes, buildworld builds includes itself, so at best this is a nop, and = it should under no circumstances be required. the key word is "should" which presumes (assumption is cockup's mother) that all files have the proper includes defined in the order of latest and greatest first so the system files are irrelevant. Now, "maybe" things are more perfect that they were... = > + this isn't enough..you have to 'make cleandir' (not 'make clean') = > + twice to remove all cruft from the obj and build dir (the first time = > + removes the obj dir completely, the second time picks up any remaining = > + cruft in the build directory itself. stale generated files in your = > + build directory can certainly cause buildworld failures. = > = > even if there are stray files after 'clean' (which there were = > not since i checked the obj tree, the function of 'depend' is = > to check the time stamps as well. = = assuming the makefiles are non-broken. some of the 'depend' targets = in the source tree are very dubious. kdump, ktrace and sysinstall are = three targets which usually refuse to rebuild generated files when = dependencies change (and 'make depend' does nothing). = = > running cleandir twice still left me after 'make all' with: = > = > makedevs.c:54: syntax error before ',' token = > makedevs.c:60: syntax error before ',' token = > *** error code 1 = = what is the relevant part of the file? I could probably pull it from the DVD archive, but I decided to run the Friday total new cvsup at 0000 GMT --the "problem" went away. I've always thought John's cvsup was infallible, but it got totally confused about a week ago when a full pull was done without cleaning out its logfiles. Just shows that you gotta play by the rules [g] ... flush one, flush both. Both buildworld and buildkernel went down without errors based on a slice from 0000 GMT 11 Oct. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message