From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-15.8 required=3.0 tests=BAYES_00,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FORGED_FROMDOMAIN,FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH,MAILING_LIST_MULTI, MENTIONS_GIT_HOSTING,SPF_HELO_NONE,SPF_PASS,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id DAAB4C07E99 for ; Mon, 5 Jul 2021 12:32:13 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id B8E46613AA for ; Mon, 5 Jul 2021 12:32:13 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230273AbhGEMet (ORCPT ); Mon, 5 Jul 2021 08:34:49 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:54862 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230188AbhGEMet (ORCPT ); Mon, 5 Jul 2021 08:34:49 -0400 Received: from mail-oi1-x235.google.com (mail-oi1-x235.google.com [IPv6:2607:f8b0:4864:20::235]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 3929DC061574 for ; Mon, 5 Jul 2021 05:32:12 -0700 (PDT) Received: by mail-oi1-x235.google.com with SMTP id h9so20605717oih.4 for ; Mon, 05 Jul 2021 05:32:12 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=from:to:cc:subject:date:message-id:mime-version :content-transfer-encoding; bh=Xo514KyQS+dsaO5cFoYPBucJn0p7vQ89OK5RTI4YUZE=; b=hK/WGpMlfKzMpdziLWTcJmPkjE/MaqTxprpnIi8SiOG2StXTK2CHVisBQV30ytHtKx XsARMHpojn9oB2PBcvDaf0AwylpASoA0ttd5qwWUbFuOvZgaHcymWPsKtbZmwijiqAMx KLWO6FHsdnwL3EY8lA4xwQw2Ltrl8WgXu573cZdUKpB9svNoA2GHmZtEfs0rUjEFj0L1 VR5ToIj3Zxf6TZFaH7M0arjkJdSnyDxGANFJyCBTvbcVxxQUIYgiNltPsNAj5GuWIjD0 Jd2pTHKqjMjHL55CvXpvEgPCrfwkxhgMqNhODr2S/Z+/uG+pAlYWpY1Yf9T/NU5AyW/2 UIIg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id:mime-version :content-transfer-encoding; bh=Xo514KyQS+dsaO5cFoYPBucJn0p7vQ89OK5RTI4YUZE=; b=LMgve14OH5DGBNbB8kN6CiFMKGzs6TpE7RBAU1Tf5bzn9gW0GkzttjOp/YSNnu41EE m+brP6ra83tGsrx2zTT7UWBV4MNc3VSYhcGBDKcMFDEUbHwq9jtwbSGNUlfVA1ZGwEyr ycE8lmohupXAfVWEabCWOIS8zx3TllceDzXknpD/k6pLOTPFXyM6YjCgxhq5S1Ch7hiI jWKgb8LNq6i9fYxwnc0oZ/IjFOyoZjejzPQbN3F47vGvCz+ChQHyJKzs66tzZc2v7b4X e9T50MZhanh4p1I/fANY58InwUbBwkyBQ4UMinj8asOoeYfLouh2H9FLvKvP7h1FVGbK qZTw== X-Gm-Message-State: AOAM530vGwcuzDS5HgmsWYfdiaQGp0C5fAu8//2/AhGEVwfsDwlHwy/S nqApscEh6qh9ylO0fK3yHlj2ejghf56VUw== X-Google-Smtp-Source: ABdhPJxaePrXQXRKNbKgO7O/vDCCw4UEi/WPlH0SyXhCCZ/wzhXwWu21AJP5HQCvdYTX5Uh525uH6A== X-Received: by 2002:aca:2802:: with SMTP id 2mr10151395oix.46.1625488331262; Mon, 05 Jul 2021 05:32:11 -0700 (PDT) Received: from localhost (fixed-187-189-187-231.totalplay.net. [187.189.187.231]) by smtp.gmail.com with ESMTPSA id z15sm2559175otj.33.2021.07.05.05.32.10 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 05 Jul 2021 05:32:10 -0700 (PDT) From: Felipe Contreras To: git@vger.kernel.org Cc: Alex Henrie , Richard Hansen , Junio C Hamano , Felipe Contreras Subject: [RFC PATCH 00/35] git update: fix broken git pull Date: Mon, 5 Jul 2021 07:31:34 -0500 Message-Id: <20210705123209.1808663-1-felipe.contreras@gmail.com> X-Mailer: git-send-email 2.32.0.36.g70aac2b1aa MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Precedence: bulk List-ID: X-Mailing-List: git@vger.kernel.org Throughout the years many big threads have been started [1] [2] [3] [4] [5] [6] because `git pull` is broken (for everyone that isn't a maintainer), and nothing has been fixed. The two main problems are: 1. The order of the parents is the opposite of what it should be 2. By default it should fail unless it's a fast-forward However, every attempt to fix these issues gets entangled in an inescapable web of interrelated problems. That's because `git pull` is in fact *two* commands pretending to be one. I have spent several days re-reading old threads looking for anything related to `git pull` and fast-forward throughout the entire history of the git mailing list, going as far back as 2008. This process has been illuminating. The amount of people that have pondered about, and even suggested, a `git update` command is striking: * Linus Torvalds: [7] * Junio C. Hamano: [8] * Richard Hansen: [9] * John Keeping: [10] And of course me [11], who not only suggested it, but implemented it as well [12]. It's not just the name, but the behavior too: `git fetch` + `git merge --ff-only` This way `git pull` can be used by maintainers to integrate pull requests, and `git update` for everyone else (normal developers) to update their local branch to its upstream. Plus it has a --merge mode that creates a merge commit, but with the parents in the right order (master to upstream, not upstream to master). Both problems are fixed by this approach. Also, a thorny issue since a long time ago has been how to properly advise users what to do in case a fast-forward is not possible. I propose a new command: `git fast-forward`. When a fast-forward is not possible `git fast-forward` fails, and shows an advice (that can be easily turned off). Therefore in fact `git update` is really `git fetch` + `git fast-forward`, and it's the latter that throws the advice. The advice also suggests `git help fast-forward` which has an explanation about what a fast-forward is, what are the options to synchronize with the remote branch (merge and rebase), and what do they look like. In addition there's a ton of fixes for `git pull` which have been sent before, including the `pull.mode` configuration, the `--merge` option, the per-branch `pullMode` configuration, `pull.mode=fast-forward`, and an improved advice message to prepare users for a future default change: The pull was not a fast-forward, in the future you will have to choose between a merge or a rebase. To quell this message you have two main options: 1. Adopt the new behavior: git config --global pull.mode fast-forward 2. Maintain the current behavior: git config --global pull.mode merge For now we will fall back to the traditional behavior: merge. For more information check "git help fast-forward". It doesn't just improve the situation for normal developers, but it also opens the possibility to fix a wart in `git pull --rebase`: Right now: git pull --merge github john (merge "github/john" into "master) git pull --rebase github john (rebase "master" onto "github/john") Since now `git update --rebase` can be used to rebase the current branch onto its upstream, that leaves `git pull --rebase` open to rebase github/john onto master, not the other way around like it currently is. Junio has suggested there's no consensus over fast-forward by default, but that's not true. Here is a non-exhaustive list of people that have agreed the default should be fast-forward only. * Linus Torvalds * Junio C Hamano * Jeff King * Jonathan Nieder * Richard Hansen * Philip Oakley * Elijah Newren * John Keeping * Ramkumar Ramachandra * Alex Henrie * W. Trevor King * Greg Troxel * Peter Kjellerstedt * Konstantin Tokarev * Robert Dailey * Vít Ondruch * Raymond E. Pasco * Jacob Keller * Felipe Contreras There's only one person against the change: Matthieu Moy. His argument back then [13] was: why ask users to merge or rebase if newcomers will not pick a rebase? To that my response is: for the same reason we ask them to do `git commit --all`: so that they eventually learn about the staging area. Just like they can do `git commit --all`, they can do `git pull --merge`. They can figure out what that means later. The people in favor of reordering the parents: * Andreas Krey * John Szakmeister * Jeremy Rosen * John Keeping And the people in favor of my old `git update`: * Damien Robert * Ping Yin * Philippe Vaucher This patch series doesn't implement all of the features my old `git update` had, but it implements most of them, most importantly--unlike `git pull`--it's not broken. Right now only `git update` without arguments works, and only if the upstream branch is configured. I have a pretty good idea of what the it should do with different arguments (as I had already implemented that), but for now that's open to discussion. If you want to give it a try: https://github.com/felipec/git/tree/fc/update [1] https://lore.kernel.org/git/200910201947.50423.trast@student.ethz.ch/ [2] https://lore.kernel.org/git/20130522115042.GA20649@inner.h.apk.li/ [3] https://lore.kernel.org/git/1377988690-23460-1-git-send-email-felipe.contreras@gmail.com/ [4] https://lore.kernel.org/git/4ay6w9i74cygt6ii1b0db7wg.1398433713382@email.android.com/ [5] https://lore.kernel.org/git/5363BB9F.40102@xiplink.com/ [6] https://lore.kernel.org/git/20201204061623.1170745-1-felipe.contreras@gmail.com/ [7] https://lore.kernel.org/git/CA+55aFzxsNxgKD1uGZQCiib+=+wCMSa0=B+Ye3Zi-u6kpz8Vrg@mail.gmail.com/ [8] https://lore.kernel.org/git/xmqqppjyhnom.fsf@gitster.dls.corp.google.com/ [9] https://lore.kernel.org/git/5228A14B.3000804@bbn.com/ [10] https://lore.kernel.org/git/20130628174252.GF2232@serenity.lan/ [11] https://lore.kernel.org/git/5366db742d494_18f9e4b308aa@nysa.notmuch/ [12] https://github.com/felipec/git/commit/61e05c9798b3c74bab8b2d47ede4c12e8e305345 [13] https://lore.kernel.org/git/vpqbo3za8r9.fsf@anie.imag.fr/ Felipe Contreras (35): merge: improve fatal fast-forward message merge: split cmd_merge() fast-forward: add new builtin doc: fast-forward: explain what it is fast-forward: add advice for novices fast-forward: make the advise configurable fast-forward: add help about merge vs. rebase update: new built-in update: add options and usage skeleton update: add --ff option update: add --merge mode commit: support for multiple MERGE_MODE merge: add --reverse-parents option update: reverse the order of parents update: fake a reverse order of parents in message update: add --rebase mode update: add mode configuation update: add per-branch mode configuration pull: cleanup autostash check pull: trivial cleanup pull: trivial whitespace style fix pull: introduce --merge option rebase: add REBASE_DEFAULT pull: move configuration fetches pull: show warning with --ff options pull: add pull.mode pull: add per-branch mode configuration pull: add pull.mode=fast-forward pull: reorganize mode conditionals pull: add diverging advice on fast-forward mode pull: improve --rebase and pull.rebase interaction pull: improve default warning pull: advice of future changes FUTURE: pull: enable ff-only mode by default !fixup FUTURE: pull: enable ff-only mode by default .gitignore | 2 + Documentation/config.txt | 4 + Documentation/config/advice.txt | 2 + Documentation/config/branch.txt | 10 ++ Documentation/config/pull.txt | 6 + Documentation/config/update.txt | 5 + Documentation/git-fast-forward.txt | 104 +++++++++++++++++ Documentation/git-pull.txt | 10 +- Documentation/git-update.txt | 47 ++++++++ Documentation/merge-options.txt | 4 + Makefile | 2 + advice.c | 15 +++ advice.h | 2 + builtin.h | 2 + builtin/commit.c | 7 +- builtin/merge.c | 59 +++++++--- builtin/pull.c | 156 +++++++++++++++++-------- builtin/update.c | 153 ++++++++++++++++++++++++ contrib/completion/git-completion.bash | 22 ++++ fmt-merge-msg.c | 21 +++- fmt-merge-msg.h | 3 +- git.c | 2 + rebase.c | 12 ++ rebase.h | 13 ++- t/t4013-diff-various.sh | 2 +- t/t5520-pull.sh | 123 +++++++++++++++++-- t/t5521-pull-options.sh | 4 +- t/t5524-pull-msg.sh | 4 +- t/t5553-set-upstream.sh | 14 +-- t/t5563-update.sh | 87 ++++++++++++++ t/t5604-clone-reference.sh | 4 +- t/t6402-merge-rename.sh | 16 +-- t/t6409-merge-subtree.sh | 6 +- t/t6417-merge-ours-theirs.sh | 10 +- t/t7600-merge.sh | 47 ++++++++ t/t7601-merge-pull-config.sh | 116 ------------------ t/t7603-merge-reduce-heads.sh | 2 +- 37 files changed, 870 insertions(+), 228 deletions(-) create mode 100644 Documentation/config/update.txt create mode 100644 Documentation/git-fast-forward.txt create mode 100644 Documentation/git-update.txt create mode 100644 builtin/update.c create mode 100755 t/t5563-update.sh -- 2.32.0.36.g70aac2b1aa