From owner-freebsd-ports@FreeBSD.ORG Sat Feb 21 12:49:00 2009 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 8857E106566B for ; Sat, 21 Feb 2009 12:49:00 +0000 (UTC) (envelope-from makc@freebsd.org) Received: from mail.issp.ac.ru (mail.issp.ac.ru [77.236.34.3]) by mx1.freebsd.org (Postfix) with ESMTP id E1C458FC17 for ; Sat, 21 Feb 2009 12:48:59 +0000 (UTC) (envelope-from makc@freebsd.org) Received: from [62.63.84.138] [62.63.84.138:43504] (HELO/EHLO luna.dio.ru, authenticated with PLAIN) by mail.issp.ac.ru with ESMTP/inet id n1LCnpNx093926 (using TLSv1/SSLv3, with cipher DHE-DSS-AES256-SHA (256 bits), verified NO) Sat, 21 Feb 2009 15:49:51 +0300 (MSK) From: Max Brazhnikov To: freebsd-ports@freebsd.org Date: Sat, 21 Feb 2009 15:48:56 +0300 User-Agent: KMail/1.11.0 (FreeBSD/7.1-STABLE; KDE/4.2.0; i386; ; ) References: <86bpsw2tbf.fsf@gmail.com> <20090221104501.06e778f1@it.buh.tecnik93.com> <86r61svw0u.fsf@gmail.com> In-Reply-To: <86r61svw0u.fsf@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200902211548.57102.makc@freebsd.org> X-Greylist: Sender succeeded SMTP AUTH authentication, not delayed by milter-greylist-3.0 (mail.issp.ac.ru [77.236.34.3]); Sat, 21 Feb 2009 15:49:51 +0300 (MSK) X-Virus-Scanned: ClamAV 0.94.1/9019/Sat Feb 21 08:23:37 2009 on mail.issp.ac.ru X-Virus-Status: Clean Cc: Anonymous Subject: Re: git snapshots, PORTVERSION, PORTEPOCH 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 Feb 2009 12:49:00 -0000 On Sat, 21 Feb 2009 15:01:37 +0300, Anonymous wrote: > Ion-Mihai Tetcu writes: > > On Sat, 21 Feb 2009 09:32:52 +0300 > > > > Anonymous wrote: > >> Would it be okay if I use commit hash in PORTVERSION and constantly > >> bump PORTEPOCH on each update? Are there any such precedents? > >> > >> %%% > >> PORTVERSION= 0.0.10${SNAPSUFFIX} > >> PORTEPOCH= 1 > >> > >> SNAPSUFFIX= .${SNAPTYPE}.${SNAPREV} > >> SNAPTYPE= git > >> SNAPREV= e09f50e > >> %%% > >> > >> Where PKGNAME would look like > >> > >> myport-0.0.10.git.e09f50e,1 > >> > >> I want to keep commit reference in place and refrain from using vague > >> dates in PORTVERSION because there can be several commits per day. And > >> for curious users I can include ChangeLog file in distfile generated > >> from git-log command. > > > > No, please don't do this. We use PORTEPOCH when there's no other way. > > OK. I can include date before commit hash and drop use of PORTEPOCH. > It would look like > > SNAPSUFFIX= .20090219.e09f50e Please, see http://www.freebsd.org/doc/en_US.ISO8859-1/books/porters- handbook/makefile-naming.html#PORTING-PKGNAME Using 'e09f50e' in PORTVERSION is not a good idea. However you can use your ${SNAPREV} in DISTVERSION. > Any other objections? > > > Use, like other ports do: > > PORTVERSION= 0.0.10 > > PORTREVISION= ${SNAPDATE} > > > > I don't see what role SNAPTYPE would have. > > SNAPTYPE isn't neccessary, just a little convenience. > > > Since you have the date, you can easily get the git magic string. > > Dates aren't atomic. That would require precise dates up to seconds and > still leave place for ambiguity. I opt to not drop git magic string if > possible else talking to upstream would be a little harder.