From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 10700 invoked by alias); 22 Feb 2011 21:23:05 -0000 Received: (qmail 10691 invoked by uid 22791); 22 Feb 2011 21:23:04 -0000 X-SWARE-Spam-Status: No, hits=-1.8 required=5.0 tests=AWL,BAYES_00,FREEMAIL_FROM,RCVD_IN_DNSWL_NONE,RFC_ABUSE_POST,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Received: from qmta13.westchester.pa.mail.comcast.net (HELO qmta13.westchester.pa.mail.comcast.net) (76.96.59.243) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Tue, 22 Feb 2011 21:22:57 +0000 Received: from omta22.westchester.pa.mail.comcast.net ([76.96.62.73]) by qmta13.westchester.pa.mail.comcast.net with comcast id B9EX1g0091ap0As5D9Nx1r; Tue, 22 Feb 2011 21:22:57 +0000 Received: from up.mrs.kithrup.com ([24.4.193.8]) by omta22.westchester.pa.mail.comcast.net with comcast id B9Nr1g0090BKwT43i9Nsg9; Tue, 22 Feb 2011 21:22:54 +0000 Subject: Re: [testsuite, build] Convert boehm-gc testsuite to DejaGnu (PR boehm-gc/11412) Mime-Version: 1.0 (Apple Message framework v1082) Content-Type: text/plain; charset=us-ascii From: Mike Stump In-Reply-To: Date: Tue, 22 Feb 2011 22:07:00 -0000 Cc: GCC Patches , Hans Boehm , Paolo Bonzini , Ben Elliston , Ralf Wildenhues Content-Transfer-Encoding: quoted-printable Message-Id: <9C6E5D0A-FCB3-4546-8920-7405261913D0@comcast.net> References: <2F36C312-AD33-470E-8169-EFE733661C01@comcast.net> To: Rainer Orth X-IsSubscribed: yes Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org X-SW-Source: 2011-02/txt/msg01484.txt.bz2 On Feb 22, 2011, at 9:09 AM, Rainer Orth wrote: > My primary issue isn't so much with the design of DejaGnu (rather its > lack of documentation), but with the current uses in GCC: duplicating > the whole per-tool code for every tool with just a few often diverging > changes isn't my idea of a maintainable code base. Yeah, happens when someone wants to fix one testsuite, but isn't given free= reign to modify others. The benefit, one change on one side can't hurt th= e other side. The downside, large scale replication. I'd support refactor= ing things... it would be a thankless job. One area in particular that I'= d love to see improved are the loops like: foreach src [lsort [find $srcdir/$subdir *_main.c]] { # If we're only testing specific files and this isn't one of them, skip= it.=20=20=20=20=20=20=20=20=20=20=20=20=20 if ![runtest_file_p $runtests $src] then { continue } compat-execute $src $sid $compat_use_alt } in the .exp testcase files. Longer term, I'd like to modify our framework = driver .exp file (those in lib) to do tests in parallel directly. I almost= had it all wired up last weekend, but ran into two problems that made me s= ad, tcl sucks and the threads people that did the proc replacement function= for quirting code into new threads, can't handle the full generality (proc= ${tool}_init ...) of tcl. The other thing that made me sad was the shear = replication of this loop in all the .exp files.