From mboxrd@z Thu Jan 1 00:00:00 1970 From: Felipe Contreras Subject: Should git-remote-hg/bzr be part of the core? Date: Sun, 11 May 2014 18:34:08 -0500 Message-ID: <537008f06ceb8_8e47492f89f@nysa.notmuch> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Cc: Richard Hansen , Torsten =?UTF-8?B?QsO2Z2Vyc2hhdXNlbg==?= , Antoine Pelisse , Christophe Simonis , Dusty Phillips , Jeff King , John Keeping To: git@vger.kernel.org, git-fc@googlegroups.com X-From: git-owner@vger.kernel.org Mon May 12 01:35:14 2014 Return-path: Envelope-to: gcvg-git-2@plane.gmane.org Received: from vger.kernel.org ([209.132.180.67]) by plane.gmane.org with esmtp (Exim 4.69) (envelope-from ) id 1WjdGy-0002in-US for gcvg-git-2@plane.gmane.org; Mon, 12 May 2014 01:35:13 +0200 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758228AbaEKXfH (ORCPT ); Sun, 11 May 2014 19:35:07 -0400 Received: from mail-oa0-f44.google.com ([209.85.219.44]:45465 "EHLO mail-oa0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758096AbaEKXeS (ORCPT ); Sun, 11 May 2014 19:34:18 -0400 Received: by mail-oa0-f44.google.com with SMTP id o6so440017oag.31 for ; Sun, 11 May 2014 16:34:17 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=date:from:to:cc:message-id:subject:mime-version:content-type :content-transfer-encoding; bh=uHxY1uS7le4p8vyR/EDwcJ4nwzbZ8ALjCPyYVupl/ec=; b=xPdRM7Je2e+KoBgnZQL02FsBF19OcSDQhmMQ4V15PU7CuabP9AL3XsXQkNo8GtD7Kd IzNdmkhuMDrPKbWZClZ1nNTEc+z2S2PSQ1X/C3iZd0vLdfRL4g9q2u09gHLHmCJJXLU2 Hg8nyHcbQilAqdNnXJBqjL2UH1sIs8Q4Qux3ci8tTMrKvyWX5jP5iMneip3Il7yMnLTS ztoOebOz7Tn+6/KgiO7SO4KvAWSubw9+MOEW///ntJoL7BJnG5xdLyhhA6bpe7h0Xzni OuI0XP7KJM4Iif9okW0uTIsVJ7BxAotgEThRnAbo1cyWtkAwcPiHD0ZfUdiWtjli76q5 Wy+Q== X-Received: by 10.182.249.18 with SMTP id yq18mr10263303obc.37.1399851257107; Sun, 11 May 2014 16:34:17 -0700 (PDT) Received: from localhost (189-211-224-40.static.axtel.net. [189.211.224.40]) by mx.google.com with ESMTPSA id js4sm42718056oeb.3.2014.05.11.16.34.14 for (version=TLSv1.2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Sun, 11 May 2014 16:34:16 -0700 (PDT) Sender: git-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: git@vger.kernel.org Archived-At: Hi, Recently Junio said he was willing to hear the opinion of other people regarding the move from contrib to the core[1]. This move is already under way, but suddenly Junio changed his mind. I have repeatedly asked him to clarify what argument changed his mind, but he hasn't done so. Hopefully he will do that in this thread, but I'll jump ahead and assume it's this one by John Keeping[2]: I do not want to end up in a situation where an update to Git is blocked by a distribution because git-remote-hg is not updated to support newer versions of Mercurial sufficiently quickly; this previously happened in Gentoo due to git-svn and meant that was stuck on 1.7.8 until 1.7.13 was released [X]. Now, I feel I addressed this argument at length, specially in this thread [3], but I'll try to provide a summary of the strongest arguments against: 1) We can make the tests optional, say 't/optional'. So if they don't pass, the build of Git is not broken. Additionally, distributions might prefer to run test-essential if they don't want to run these optional tests. 2) There's already continuous integration builds[4] to make sure all the possible incoming changes in Mercurial are detected early on. 3) There has been a *single* case where a new Mercurial (3.0) broke things. This is due to the fact of how I implemented certain functionality due to limitations in Mercurial's API. Mercurial authors can be contacted to improve the API and minimize the possibility of it happening again. Given these arguments, I don't see how moving git-remote-hg to the core could possibly cause any problems, and I don't understand why Junio would think otherwise. Even if such a breakage causes a problem to the whole Git, it would make sense to demote these tools *when* such a problem comes (which I argue it won't). Does it make sense to you that you get thrown in jail for a crime you haven't committed merely because someone thinks it's likely you will? Given the huge amount of work I've put in these remote helpers, and the fact that Junio said since day 1 he wanted these in the core[5] (and I was operating under that assumption), I think the demotion back to the contrib area (and therefore out-of-tree) should be made carefully, and not from one day to he next as it happened. Thoughts? [1] http://article.gmane.org/gmane.comp.version-control.git/248676 [2] http://article.gmane.org/gmane.comp.version-control.git/248167 [3] http://article.gmane.org/gmane.comp.version-control.git/248683 [4] https://travis-ci.org/felipec/git [5] http://article.gmane.org/gmane.comp.version-control.git/220277 -- Felipe Contreras