From owner-freebsd-perl@FreeBSD.ORG Wed Nov 5 16:10:50 2008 Return-Path: Delivered-To: perl@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id EE4761065677; Wed, 5 Nov 2008 16:10:50 +0000 (UTC) (envelope-from leeym@leeym.com) Received: from hs-out-0708.google.com (hs-out-0708.google.com [64.233.178.246]) by mx1.freebsd.org (Postfix) with ESMTP id 8AE0C8FC26; Wed, 5 Nov 2008 16:10:50 +0000 (UTC) (envelope-from leeym@leeym.com) Received: by hs-out-0708.google.com with SMTP id 54so38018hsz.11 for ; Wed, 05 Nov 2008 08:10:49 -0800 (PST) Received: by 10.90.25.11 with SMTP id 11mr808827agy.57.1225901449610; Wed, 05 Nov 2008 08:10:49 -0800 (PST) Received: by 10.100.189.2 with HTTP; Wed, 5 Nov 2008 08:10:49 -0800 (PST) Message-ID: <759236930811050810t2150d000w741b8e2b5e7b9aae@mail.gmail.com> Date: Wed, 5 Nov 2008 08:10:49 -0800 From: "Yen-Ming Lee" To: "Anton Berezin" In-Reply-To: <20081103125035.GA66067@heechee.tobez.org> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Content-Disposition: inline References: <200810181406.m9IE6jW8067588@freefall.freebsd.org> <759236930810180813r415eaae2of299ebed951632a8@mail.gmail.com> <20081103125035.GA66067@heechee.tobez.org> Cc: perl@freebsd.org Subject: Re: ports/128187: [PATCH] devel/p5-Parse-RecDescent: simplfy dependency X-BeenThere: freebsd-perl@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: maintainer of a number of perl-related ports List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 05 Nov 2008 16:10:51 -0000 2008/11/3 Anton Berezin : > Well, until we have a bit more automatic and a bit more comprehensive > solution, don't you think that we should consider such things on a > case-by-case basis? I understand your arguments, and I have presented my > own arguments, sometimes agreeing with you, sometimes not, but this > discussion is larger than this particular PR. > > In *this* particular case, do you have a good reason of dropping > Text::Balanced from the BUILD_DEPENDS? I don't see any, whatever way I > slice it and dice it. I use conditional dependency because it is the core module for 5.7.3, and I remove it from BUILD_DEPENDS since Parse::RecDescent can build without Text::Balanced. It will complain, but just complain. In *this* particular case, conditional dependency is my main purpose, and I'm fine if you want to keep it in BUILD_DEPENDS. >> http://lists.freebsd.org/pipermail/freebsd-perl/2008-February/001745.html >> http://lists.freebsd.org/pipermail/freebsd-perl/2008-February/001748.html >> >> 2. (I'm doing so now) put 'requires' in RUN_DEPENDS, and put something >> in BUILD_DEPENDS if really needed. Also use conditional dependencies >> to avoid dual-life modules. It complains during the configure stage, >> but it works well. > > "Really needed" is in the eye of the beholder. You know that Makefile.PL > bitching is harmless, I know it is harmless, but a person installing the > module does not. In my opinion, "really needed" means if we don't have that module installed, Makefile.PL will download it from CPAN or do whatever we don't want. If it just complains, never mind, we will install it later when we 'make install'. If you and all the perl@ developers prefer solution 1, to install all the modules so Makefile.PL will never complain, no problem, just put it in porters handbook, in case some users argue that we install unnecessary modules, I can show them that it's our conclusion. We can still review them case by case, but in solution 1, the Makefile should look like this: RUN_DEPENDS= the ones needed for run BUILD_DEPENDS= ${RUN_DEPENDS} BUILD_DEPENDS+= the ones needed for tests (if any) right? >> http://lists.freebsd.org/pipermail/freebsd-perl/2008-July/001900.html >> >> 3. use PERL_DEPENDS, but we still need to find out the relationship >> between the minimum version of the required module and the package >> version fulfill that minimum version of the module. In fact I >> encounter this problem again recently: Net::Amazon::S3-0.45 requires >> IO::File-0.14, which is from IO, but obviously not IO-0.14 > > Yes, this is a drawback of our proposed solution. I still think it is > better to rely on distributions and not on individual modules, though. Since it's the only drawback, and we think it's acceptable, why don't we just do it ? >> http://lists.freebsd.org/pipermail/freebsd-perl/2008-April/001834.html >> >> We discussed about it several times, please make a decision this time >> and put it somewhere in the porters handbook. > > Hey, *who* is it you want to make a decision? > > Cheers, > \Anton. At least not me. You are the better candidate, the core perl ports maintainer. :-) Regards, -- Yen-Ming Lee