From owner-freebsd-ports@FreeBSD.ORG Sat Jun 21 01:20:29 2008 Return-Path: Delivered-To: freebsd-ports@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 96630106566C for ; Sat, 21 Jun 2008 01:20:29 +0000 (UTC) (envelope-from bf2006a@yahoo.com) Received: from web39101.mail.mud.yahoo.com (web39101.mail.mud.yahoo.com [209.191.86.252]) by mx1.freebsd.org (Postfix) with SMTP id 248268FC18 for ; Sat, 21 Jun 2008 01:20:29 +0000 (UTC) (envelope-from bf2006a@yahoo.com) Received: (qmail 90331 invoked by uid 60001); 21 Jun 2008 00:53:47 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding:Message-ID; b=YBd2UqcGUVTE0NV2xy5R8agnE3obKCXEFaKRrGwU9fuZ/bPjpY+d+a7tUm0Jl6qPkU5roFp+mIJ4rzvUbtUCAjOovLY8De8Ger0bc/wuKQLZ97cLBmfsyahfPV/t3mWHEWoiuD2UyBd1SFftWiKb6YTVDUXgK57jBuJSFZuYlg0=; X-YMail-OSG: 8ukoMUsVM1l7KHcmXUWsFqIcCCVvQibKWdwihVCySD4Fcv_FQz_GATunQpW95JRQ4pwZ8IpTetXScvypaknWwtR_EAcTBm6BJlG5dw-- Received: from [78.47.29.82] by web39101.mail.mud.yahoo.com via HTTP; Fri, 20 Jun 2008 17:53:47 PDT Date: Fri, 20 Jun 2008 17:53:47 -0700 (PDT) From: bf To: freebsd-ports@freebsd.org MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Message-ID: <447581.87729.qm@web39101.mail.mud.yahoo.com> Subject: Re: regression-test X-BeenThere: freebsd-ports@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Porting software to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 21 Jun 2008 01:20:29 -0000 I'd agree with Thierry. Ideally, you wouldn't want to even build a piece of software that didn't work right, let alone install it. But once you admit the possibility that mistakes may have been made, and recognize that tests are desirable, then *conceptually* you want to test a port in circumstances as close as possible to those in which it will be used -- i.e., after it's been installed. But pragmatically, it really comes down to the tests that will be implemented. If this regression-test target is largely there to allow bundled testing targets to be run from the port Makefile, and these bundled tests are written with a port that is built, but not installed in mind, then run them before installation. Otherwise, afterwards. It's going to be difficult to get people to rewrite test suites just to fit someone's debatable notion of when tests should be made -- we already have enough trouble keeping up with other changes, and we have many unmaintained ports. b.