From owner-freebsd-hackers Thu Oct 31 06:29:19 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id GAA09473 for hackers-outgoing; Thu, 31 Oct 1996 06:29:19 -0800 (PST) Received: from covina.lightside.com (covina.lightside.com [207.67.176.1]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id GAA09463 for ; Thu, 31 Oct 1996 06:29:16 -0800 (PST) Received: from localhost (jehamby@localhost) by covina.lightside.com (8.8.2/8.8.2) with SMTP id GAA01736; Thu, 31 Oct 1996 06:27:59 -0800 (PST) Date: Thu, 31 Oct 1996 06:27:50 -0800 (PST) From: Jake Hamby To: Terry Lambert cc: Jonathan Lemon , joerg_wunsch@uriah.heep.sax.de, freebsd-hackers@freebsd.org, pgiffuni@fps.biblos.unal.edu.co Subject: Re: POSIX Conformance (Unanswered in "questions" so I forwarded...) In-Reply-To: <199610310002.RAA24354@phaeton.artisoft.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk On Wed, 30 Oct 1996, Terry Lambert wrote: > > Terry Lambert writes: > > > Meanwhile, TET is a completely seperate piece of software necessary > > > for running TET-hosted test suites... like NIST/PCTS. I've already > > > suggested that someone grab the TET off of the X/Open server (where it > > > is available for anonymous FTP) and check it into the FreeBSD ports > > > tree. > > > > Which one? > > > > - TET3.0a - unavailable to non-paying customers at the moment > > - dTET2.3 - "distributed TET" > > - eTET1.10.3 - "extended TET" > > - TET1.10 - The release version > > > > Without knowing which version the NIST code requires, I'd be a little > > leery of just going in and attempting a port. > > Why? A POSIX-compliant platform is required to run all of them > without changes. > > Seems like a good test-by-fire to me. > > I believe I ran under TET1.10 at Novell for the UnixWare 2.x testing. > > I used dTET2.3 under OpenBSD for NIST/PCTSthe other day I didn't have any trouble compiling TET1.10 under FreeBSD. The makefiles are allowed to be modified, and the only thing I needed to add was a define for NSIG (which is not defined in /usr/src/sys/signal.h if _POSIX_SOURCE is also defined). I also changed the signal numbering in the shell script Makefile, as described in the installation documentation. Note that it didn't test this too thoroughly, other than successfully running it with the demo testbed. TET is a very small program (208k tar archive), so there is no reason not to make a port out of it, as Terry suggests. It may be useful for testing our conformance to other standards (like X11) besides POSIX. Also, our libc complains about the use of gets(), but I didn't notice any problem running the test. Now that our linker complains about the use of gets(), though, we should probably comment out the run-time warning since it might affect the POSIX test results. Finally, I should note that GCC detected a genuine bug in the TET source code. I don't know if this has been fixed in the other versions or the 3.0 version (which is only available to members now). It is in line 477 (the last line) of journal.c, whose function returns the address of a local variable. > More information: > > http://www.nist.gov/itl/div897/pubs/fip151-2.htm This, like most of the documents on the NIST and NCSL WWW sites is not very useful! Though what else I would expect from a government agency, I don't know... -- Jake