From: Junio C Hamano Subject: Re: [PATCH/RFC] gitweb: Great subroutines renaming Date: Sun, 13 Aug 2006 17:11:30 -0700 Message-ID: <7vejvki3bh.fsf@assigned-by-dhcp.cox.net> References: <200608071626.52655.jnareb@gmail.com> <7v7j1dnzo5.fsf@assigned-by-dhcp.cox.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: git@vger.kernel.org X-From: git-owner@vger.kernel.org Mon Aug 14 02:11:41 2006 Return-path: Envelope-to: gcvg-git@gmane.org Received: from vger.kernel.org ([209.132.176.167]) by ciao.gmane.org with esmtp (Exim 4.43) id 1GCQ3M-0004Ji-P0 for gcvg-git@gmane.org; Mon, 14 Aug 2006 02:11:37 +0200 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751753AbWHNALd (ORCPT ); Sun, 13 Aug 2006 20:11:33 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751754AbWHNALd (ORCPT ); Sun, 13 Aug 2006 20:11:33 -0400 Received: from fed1rmmtao06.cox.net ([68.230.241.33]:1437 "EHLO fed1rmmtao06.cox.net") by vger.kernel.org with ESMTP id S1751753AbWHNALc (ORCPT ); Sun, 13 Aug 2006 20:11:32 -0400 Received: from assigned-by-dhcp.cox.net ([68.4.5.203]) by fed1rmmtao06.cox.net (InterMail vM.6.01.06.01 201-2131-130-101-20060113) with ESMTP id <20060814001131.KUQF6235.fed1rmmtao06.cox.net@assigned-by-dhcp.cox.net>; Sun, 13 Aug 2006 20:11:31 -0400 To: Jakub Narebski In-Reply-To: (Jakub Narebski's message of "Sun, 13 Aug 2006 11:29:00 +0200") User-Agent: Gnus/5.110006 (No Gnus v0.6) Emacs/21.4 (gnu/linux) Sender: git-owner@vger.kernel.org Precedence: bulk X-Mailing-List: git@vger.kernel.org Archived-At: Jakub Narebski writes: > Junio C Hamano wrote: > >> I'll be pushing out 1.4.2 this weekend, and then moving all the >> gitweb stuff pending in "next" to "master" after that. >> >> Let's have the rename immediately on top of it first, then >> continue cleaning up after that. > > So after 1.4.2 gitweb patches should be based on next, or on master? Well, since "git diff master next -- gitweb/" reports empty, the question should not matter at this moment ;-). Here are my preferences, as a general guideline: * If it is a general clean-up (e.g. changing a function signature declared in cache.h to tighten constness), I would prefer two patches; one to apply on "master", and another one created by assuming: - your first patch for "master" is accepted; - resulting "master" (with your first patch applied) will be merged into "next"; Then the other patch is to clean-up the remaining stuff introduced between "master" and "next" (e.g. the function whose signature you changed in the first patch may have more call sites in "next" you may need to fix them up). If you know the specific topic I internally have (you can tell them by looking at the merge commits on "next"), separating the remainder patch into one per topic would be even nicer, but that is probably too much work to ask from contributors. Most of the time I can do this split myself, though, so it is Ok to send a patch to fix things up only in "master", and have me find out that it breaks "next", at that time I may fix things up myself similarly or I may ask you to send in a separete patch in similar spirit to fix up things in "next". * If it is a new series, I would prefer things be based on "master", unless you use some feature (either internal or external) that only exists in "next" or in "pu". * If it is a follow-up on stuff still in "next", obviously basing the patch on "master" is not possible. * If you have independent fix even though we have stuff that touch the same general area in "next", it is very much preferable if it applies to "master". A critical fix for "next" that does not apply to "master" is taking the fix hostage to whatever is in "next". * If you need to base on "next", telling me why (i.e. "this uses such and such function which are not in master") helps me identify which topic I internally have to base a topic to house your series in. * If you need to base on "pu", think twice. You will be taken hostage by whatever you are basing on (this is the same if you base on something in "next", but "pu" is for stuff of even more dubious merits than "next"). On the other hand, your new series may turn out to be the killer application for the feature that was earlier felt not-so-useful and help it (and your series) merged to "next".