# Intro `git replay` has initially been developed entirely by Elijah Newren mostly last year (2022) at: https://github.com/newren/git/commits/replay I took over a few months ago to polish and upstream it as GitLab is interested in replacing libgit2, and for that purpose needs a command to do server side (so without using a worktree) rebases, cherry-picks and reverts. I reduced the number of commits and features in this first patch series, compared to what Elijah already developed. Especially I stopped short of replaying merge commits and replaying interactively. These and other features might be upstreamed in the future after this patch series has graduated. The focus in this series is to make it a good plumbing command that can already be used server side and that replaces the "fast-rebase" test-tool command. So things to make it easier to use on the command line, and more advanced features (like replaying merges) are left out. # Content of this cover letter The "Quick Overview" and "Reasons for diverging from cherry-pick & rebase" sections just below are describing the purpose of the new command in the big scheme of things. They are taken from Elijah's design notes (https://github.com/newren/git/blob/replay/replay-design-notes.txt) and describe what we want this command to become and the reasons for that, not what the command is after only this patch series. Also these design notes were written at least one year ago, so parts of those 2 sections are not true anymore. I have added Phillip Wood's or Felipe Contreras' notes (thanks to them) where that's the case, but some now flawed parts may have missed. After these two sections, starting with the "Important limitations" section, you will find sections describing what is actually in this patch series. More interesting material is available in Elijah's design notes like an "Intro via examples" (https://github.com/newren/git/blob/replay/replay-design-notes.txt#L37-L132), a discussion about "Preserving topology, replaying merges" (https://github.com/newren/git/blob/replay/replay-design-notes.txt#L264-L341) and a "Current status" section describing Elijah's work (https://github.com/newren/git/blob/replay/replay-design-notes.txt#L344-L392) before I started working on upstreaming it. I have not included this material here though, as the documentation added by this patch series for the `git replay` command already includes an "EXAMPLES" section, and other sections of Elijah's design notes might not be interesting for now. Also this cover letter is already pretty long. But reviewers can refer to the links above if they think it can help. # Quick Overview (from Elijah's design notes) `git replay`, at a basic level, can perhaps be thought of as a "default-to-dry-run rebase" -- meaning no updates to the working tree, or to the index, or to any references. However, it differs from rebase in that it: * Works for branches that aren't checked out * Works in a bare repository * Can replay multiple branches simultaneously (with or without common history in the range being replayed) * Preserves relative topology by default (merges are replayed too in Elijah's original work, not in this series) * Focuses on performance * Has several altered defaults as a result of the above I sometimes think of `git replay` as "fast-replay", a patch-based analogue to the snapshot-based fast-export & fast-import tools. # Reasons for diverging from cherry-pick & rebase (from Elijah's design notes) There are multiple reasons to diverge from the defaults in cherry-pick and rebase. * Server side needs * Both cherry-pick and rebase, via the sequencer, are heavily tied to updating the working tree, index, some refs, and a lot of control files with every commit replayed, and invoke a mess of hooks[1] that might be hard to avoid for backward compatibility reasons (at least, that's been brought up a few times on the list). * cherry-pick and rebase both fork various subprocesses unnecessarily, but somewhat intrinsically in part to ensure the same hooks are called that old scripted implementations would have called. Note: since 356ee4659bb (sequencer: try to commit without forking 'git commit', 2017-11-24) cherry-pick and rebase do not fork subprocesses other than hooks for the cases covered by this patch series (i.e. they do not fork "git commit" for simple picks). * "Dry run" behavior, where there are no updates to worktree, index, or even refs might be important. * Should not assume users only want to operate on HEAD (see next section) * Decapitate HEAD-centric assumptions * cherry-pick forces commits to be played on top of HEAD; inflexible. * rebase assumes the range of commits to be replayed is upstream..HEAD by default, though it allows one to replay upstream..otherbranch -- but it still forcibly and needlessly checks out 'otherbranch' before starting to replay things. Note: since 767a9c417eb (rebase -i: stop checking out the tip of the branch to rebase, 2020-01-24) it's not true that rebase forcibly and needlessly checks out 'otherbranch'. * Assuming HEAD is involved severely limits replaying multiple (possibly divergent) branches. Note: since 89fc0b53fdb (rebase: update refs from 'update-ref' commands, 2022-07-19) the sequencer can update multiple branches. The issue with divergent branch is with command line arguments and the todo list generation rather than the capabilities of the sequencer. * Once you stop assuming HEAD has a certain meaning, there's not much reason to have two separate commands anymore (except for the funny extra not-necessarily-compatible options both have gained over time). * (Micro issue: Assuming HEAD is involved also makes it harder for new users to learn what rebase means and does; it makes command lines hard to parse. Not sure I want to harp on this too much, as I have a suspicion I might be creating a tool for experts with complicated use cases, but it's a minor quibble.) * Performance * jj is slaughtering us on rebase speed[2]. I would like us to become competitive. (I dropped a few comments in the link at [2] about why git is currently so bad.) * From [3], there was a simple 4-patch series in linux.git that took 53 seconds to rebase. Switching to ort dropped it to 16 seconds. While that sounds great, only 11 *milliseconds* were needed to do the actual merges. That means almost *all* the time (>99%) was overhead! Big offenders: * --reapply-cherry-picks should be the default * can_fast_forward() should be ripped out, and perhaps other extraneous revision walks Note: d42c9ffa0f (rebase: factor out branch_base calculation, 2022-10-17) might already deal with that (according to Felipe Contreras). * avoid updating working tree, index, refs, reflogs, and control structures except when needed (e.g. hitting a conflict, or operation finished) * Other performance ideas (mostly for future work, not in this series) * single-file control structures instead of directory of files (when doing interactive things which is in Elijah's original work, but not in this series) * avoid forking subprocesses unless explicitly requested (e.g. --exec, --strategy, --run-hooks). For example, definitely do not invoke `git commit` or `git merge`. * Sanitize hooks: * dispense with all per-commit hooks for sure (pre-commit, post-commit, post-checkout). * pre-rebase also seems to assume exactly 1 ref is written, and invoking it repeatedly would be stupid. Plus, it's specific to "rebase". So...ignore? (Stolee's --ref-update option for rebase probably broke the pre-rebase assumptions already...) * post-rewrite hook might make sense, but fast-import got exempted, and I think of replay like a patch-based analogue to the snapshot-based fast-import. * When not running server side, resolve conflicts in a sparse-cone sparse-index worktree to reduce number of files written to a working tree. (See below as well.) * [High risk of possible premature optimization] Avoid large numbers of newly created loose objects, when replaying large numbers of commits. Two possibilities: (1) Consider using tmp-objdir and pack objects from the tmp-objdir at end of exercise, (2) Lift code from git-fast-import to immediately stuff new objects into a pack? * Multiple branches and non-checked out branches * The ability to operate on non-checked out branches also implies that we should generally be able to replay when in a dirty working tree (exception being when we expect to update HEAD and any of the dirty files is one that needs to be updated by the replay). * Also, if we are operating locally on a non-checked out branch and hit a conflict, we should have a way to resolve the conflict without messing with the user's work on their current branch. (This is not is this patch series though.) * Idea: new worktree with sparse cone + sparse index checkout, containing only files in the root directory, and whatever is necessary to get the conflicts * Companion to above idea: control structures should be written to $GIT_COMMON_DIR/replay-${worktree}, so users can have multiple replay sessions, and so we know which worktrees are associated with which replay operations. - [1] https://lore.kernel.org/git/pull.749.v3.git.git.1586044818132.gitgitgadget@xxxxxxxxx/ - [2] https://github.com/martinvonz/jj/discussions/49 - [3] https://lore.kernel.org/git/CABPp-BE48=97k_3tnNqXPjSEfA163F8hoE+HY0Zvz1SWB2B8EA@xxxxxxxxxxxxxx/ # Important limitations * The code exits with code 1 if there are any conflict. No resumability. No nice output. No interactivity. No special exit code depending on the reason. * When a commit becomes empty as it is replayed, it is still replayed as an empty commit, instead of being dropped. * No replaying merges, nor root commits. Only regular commits. * Signed commits are not properly handled. It's not clear what to do to such commits when replaying on the server side. * Notes associated with replayed commits are not updated nor duplicated. (Thanks to Phillip Wood for noticing.) # Commit overview * 1/15 t6429: remove switching aspects of fast-rebase New preparatory commit to make it easier to later replace the fast-rebase test-tool by `git replay` without breaking existing tests. * 2/15 replay: introduce new builtin This creates a minimal `git replay` command by moving the code from the `fast-rebase` test helper from `t/helper/` into `builtin/` and doing some renames and a few other needed changes. (No change on this commit since v1.) * - 3/15 replay: start using parse_options API - 4/15 replay: die() instead of failing assert() - 5/15 replay: introduce pick_regular_commit() - 6/15 replay: don't simplify history - 7/15 replay: add an important FIXME comment about gpg signing - 8/15 replay: remove progress and info output - 9/15 replay: remove HEAD related sanity check These slowly change the command to make it behave more like a regular commands and to start cleaning up its output. (No change on these commits since v1.) * 10/15 replay: make it a minimal server side command After the cleaning up in previous commits, it's now time to radically change the way it works by stopping it to do ref updates, to update the index and worktree, to consider HEAD as special. Instead just make it output commands that should be passed to `git update-ref --stdin`. (No change on this commit since v1.) * 11/15 replay: use standard revision ranges Start addind new interesting features and also documentation and tests, as the interface of the command is cristalizing into its final form. Since v1, new tests to check that the command works well on bare repo and exits with code 1 in case of conflict have been added. The doc also now says that the new command is experimental. * 12/15 replay: disallow revision specific options and pathspecs New commit to disallow for now revision specific options and pathspecs that are allowed and eaten by setup_revisions(), as it's not clear if all of these extra features are really needed, and anyway they would require tests and doc. So we leave them for future improvements. * - 13/15 replay: add --advance or 'cherry-pick' mode - 14/15 replay: add --contained to rebase contained branches New commits, but extracted from a previous commit (that added both --advance and --contained at the same time) without any code change except that a few test cases for bare repos were added. * 15/15 replay: stop assuming replayed branches do not diverge This adds another interesting feature, as well as related documentation and tests. (Since v1 a few test cases for bare repos were added.) # Notes about `fast-rebase`, tests and documentation The `fast-rebase` test-tool helper was developed by Elijah to experiment with a rebasing tool that would be developed from scratch based on his merge-ort work, could be used to test that merge-ort work, and would not have the speed and interface limitations of `git rebase` or `git cherry-pick`. This `fast-rebase` helper was used before this series in: t6429-merge-sequence-rename-caching.sh So when `git replay` is created from `fast-rebase` in patch 2/15, the t6429 test script is also converted to use `git replay`. This ensures that `git replay` doesn't break too badly during the first 10 patches in this patch series. Tests and documentation are introduced specifically for `git replay` only in 11/15 and later patches as it doesn't make much sense to document and test behavior that we know is going to change soon. So it's only when the command is crystalizing towards its final form that we start documenting and testing it. # Possibly controversial issues * bare or not bare: this series works towards a plumbing command with the end goal of it being usable and used first on bare repos, contrary to existing commands like `git rebase` and `git cherry-pick`. The tests check that the command works on both bare and non-bare repo though. * exit status: a successful, non-conflicted replay exits with code 0. When the replay has conflicts, the exit status is 1. If the replay is not able to complete (or start) due to some kind of error, the exit status is something other than 0 or 1. There are a few tests checking that. It has been suggested in an internal review that conflicts might want to get a more specific error code as an error code of 1 might be quite easy to return by accident. It doesn't seem to me from their docs (which might want to be improved, I didn't look at the code) that other commands like `git merge` and `git rebase` exit with a special error code in case of conflict. * make worktree and index changes optional: commit 10/15 stops updating the index and worktree, but it might be better especially for cli users to make that optional. The issue is that this would make the command more complex while we are developing a number of important features so that the command can be used on bare repos. It seems that this should rather be done in an iterative improvement after the important features have landed. * when and where to add tests and docs: although t6429 has tests that are changed to use the new command instead of the fast-rebase test-tool command as soon as the former is introduced, there is no specific test script and no doc for the new command until commit 11/15 when standard revision ranges are used. This is done to avoid churn in tests and docs while the final form of the command hasn't crystalized enough. Adding tests and doc at this point makes this commit quite big and possibly more difficult to review than if they were in separate commits though. On the other hand when tests and docs are added in specific commits some reviewers say it would be better to introduce them when the related changes are made. * --advance and --contained: these two advanced options might not belong to this first series and could perhaps be added in a followup series in separate commits. On the other hand the code for --contained seems involved with the code of --advance and it's nice to see soon that git replay can indeed do cherry-picking and rebase many refs at once, and this way fullfil these parts of its promise. * replaying diverging branches: 15/15 the last patch in the series, which allow replaying diverging branches, can be seen as a fundamental fix or alternatively as adding an interesting feature. So it's debatable if it should be in its own patch along with its own tests as in this series, or if it should be merged into a previous patch and which one. * only 2 patches: this patch series can be seen from a high level point of view as 1) introducing the new `git replay` command, and 2) using `git replay` to replace, and get rid of, the fast-rebase test-tool command. The fact that not much of the original fast-rebase code and interface is left would agree with that point of view. On the other hand, fast-rebase can also be seen as a first iteration towards `git replay`. So it can also make sense to see how `git replay` evolved from it. # Changes between v1 and v2 Thanks to all the reviewers on the mailing list or during a Review Club! Especially, thanks to Elijah, Dscho, Phillip Wood, Derrick Stolee, Patrick Steinhardt, Junio, Felipe Contreras, Calvin Wan and all the Review Club participants. The Review Club notes (see those on 2023-04-18) are available on: https://docs.google.com/document/d/14L8BAumGTpsXpjDY8VzZ4rRtpAjuGrFSRqn3stCuS_w/ Now the changes between v1 and v2 are: * The patch series has been rebased on top of current master, so that conflicts with master are resolved. This required replacing calls to functions like get_oid() with repo_get_oid(), as well as changing some `#include "*.h"` to use different header files. This unfortunately makes the range-diff between v1 and v2 a bit more difficult to read than we would like. * Patch 1/15 is a new patch. It has been introduced to rework t6429, so that this test script doesn't break when conflict output is removed later in the series. This allowed us to remove the old 9/14 patch ("replay: very coarse worktree updating"). * The old 9/14 patch ("replay: very coarse worktree updating") was removed. This is to make sure that the new command can be used on bare repos. This enables us to focus first on making sure the new command a good plumbing command that works well on bare repos. * The old 12/14 patch ("replay: introduce guess_new_base()") was removed. This is also to focus first on making sure the new command is a good plumbing command. * The old 13/14 patch ("replay: add different modes") has been split into 2 patches: patch 13/15 ("replay: add --advance or 'cherry-pick' mode") and patch 14/15 ("replay: add --contained to rebase contained branches"). * Patch 12/15 ("replay: disallow revision specific options and pathspecs") is a new patch that disallow using revision specific options that would be allowed and eaten by setup_revisions(). Even if it would be very nice to have some of these options in the future, not all these options make sense for the new command, and it might require improving the setup_revisions() mechanism (as well as adding tests and doc) to cleanly allow those we want, but that is outside the scope of this first patch series. While at it, this patch also disallow pathspecs too. It would be nice to have them in the future, but for now it is not necessary and would require work (at least for tests and doc) that can be done later. * Patches 11/15 to 15/15 include a number of new tests to check that the command works well on bare repos, and in case of 11/15 to check that exit code in case of conflict is 1. * Patch 11/15, in which the doc for the `git replay` command is introduced, now says that the new command is experimental. * Note that a segfault that was found by Derrick Stolee in the v1 is fixed by 000c4ceca7 (merge-ort: fix calling merge_finalize() with no intermediate merge, 2023-04-22) merged to master by 3927312601 (Merge branch 'en/ort-finalize-after-0-merges-fix', 2023-05-02). Christian Couder (1): replay: disallow revision specific options and pathspecs Elijah Newren (14): t6429: remove switching aspects of fast-rebase replay: introduce new builtin replay: start using parse_options API replay: die() instead of failing assert() replay: introduce pick_regular_commit() replay: don't simplify history replay: add an important FIXME comment about gpg signing replay: remove progress and info output replay: remove HEAD related sanity check replay: make it a minimal server side command replay: use standard revision ranges replay: add --advance or 'cherry-pick' mode replay: add --contained to rebase contained branches replay: stop assuming replayed branches do not diverge .gitignore | 1 + Documentation/git-replay.txt | 125 +++++++ Makefile | 2 +- builtin.h | 1 + builtin/replay.c | 431 +++++++++++++++++++++++ command-list.txt | 1 + git.c | 1 + t/helper/test-fast-rebase.c | 240 ------------- t/helper/test-tool.c | 1 - t/helper/test-tool.h | 1 - t/t3650-replay-basics.sh | 206 +++++++++++ t/t6429-merge-sequence-rename-caching.sh | 45 +-- 12 files changed, 792 insertions(+), 263 deletions(-) create mode 100644 Documentation/git-replay.txt create mode 100644 builtin/replay.c delete mode 100644 t/helper/test-fast-rebase.c create mode 100755 t/t3650-replay-basics.sh -- 2.40.1.491.gdff9a222ea
