Git Source Code Mirror - This is a publish-only repository and all pull requests are ignored. Please follow Documentation/SubmittingPatches procedure for any of your improvements. https://git-scm.com/
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
git/sequencer.c

6384 lines
176 KiB

#include "cache.h"
#include "config.h"
#include "lockfile.h"
#include "dir.h"
#include "object-store.h"
#include "object.h"
#include "commit.h"
#include "sequencer.h"
#include "tag.h"
#include "run-command.h"
#include "hook.h"
#include "exec-cmd.h"
#include "utf8.h"
#include "cache-tree.h"
#include "diff.h"
#include "revision.h"
#include "rerere.h"
#include "merge-ort.h"
#include "merge-ort-wrappers.h"
#include "refs.h"
#include "strvec.h"
#include "quote.h"
#include "trailer.h"
#include "log-tree.h"
#include "wt-status.h"
#include "hashmap.h"
#include "notes-utils.h"
#include "sigchain.h"
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
#include "unpack-trees.h"
#include "worktree.h"
#include "oidmap.h"
#include "oidset.h"
#include "commit-slab.h"
#include "alias.h"
#include "commit-reach.h"
#include "rebase-interactive.h"
#include "reset.h"
rebase: add --update-refs option When working on a large feature, it can be helpful to break that feature into multiple smaller parts that become reviewed in sequence. During development or during review, a change to one part of the feature could affect multiple of these parts. An interactive rebase can help adjust the multi-part "story" of the branch. However, if there are branches tracking the different parts of the feature, then rebasing the entire list of commits can create commits not reachable from those "sub branches". It can take a manual step to update those branches. Add a new --update-refs option to 'git rebase -i' that adds 'update-ref <ref>' steps to the todo file whenever a commit that is being rebased is decorated with that <ref>. At the very end, the rebase process updates all of the listed refs to the values stored during the rebase operation. Be sure to iterate after any squashing or fixups are placed. Update the branch only after those squashes and fixups are complete. This allows a --fixup commit at the tip of the feature to apply correctly to the sub branch, even if it is fixing up the most-recent commit in that part. This change update the documentation and builtin to accept the --update-refs option as well as updating the todo file with the 'update-ref' commands. Tests are added to ensure that these todo commands are added in the correct locations. This change does _not_ include the actual behavior of tracking the updated refs and writing the new ref values at the end of the rebase process. That is deferred to a later change. Signed-off-by: Derrick Stolee <derrickstolee@github.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 months ago
#include "branch.h"
#include "log-tree.h"
#define GIT_REFLOG_ACTION "GIT_REFLOG_ACTION"
static const char sign_off_header[] = "Signed-off-by: ";
static const char cherry_picked_prefix[] = "(cherry picked from commit ";
GIT_PATH_FUNC(git_path_commit_editmsg, "COMMIT_EDITMSG")
static GIT_PATH_FUNC(git_path_seq_dir, "sequencer")
static GIT_PATH_FUNC(git_path_todo_file, "sequencer/todo")
static GIT_PATH_FUNC(git_path_opts_file, "sequencer/opts")
static GIT_PATH_FUNC(git_path_head_file, "sequencer/head")
static GIT_PATH_FUNC(git_path_abort_safety_file, "sequencer/abort-safety")
memoize common git-path "constant" files One of the most common uses of git_path() is to pass a constant, like git_path("MERGE_MSG"). This has two drawbacks: 1. The return value is a static buffer, and the lifetime is dependent on other calls to git_path, etc. 2. There's no compile-time checking of the pathname. This is OK for a one-off (after all, we have to spell it correctly at least once), but many of these constant strings appear throughout the code. This patch introduces a series of functions to "memoize" these strings, which are essentially globals for the lifetime of the program. We compute the value once, take ownership of the buffer, and return the cached value for subsequent calls. cache.h provides a helper macro for defining these functions as one-liners, and defines a few common ones for global use. Using a macro is a little bit gross, but it does nicely document the purpose of the functions. If we need to touch them all later (e.g., because we learned how to change the git_dir variable at runtime, and need to invalidate all of the stored values), it will be much easier to have the complete list. Note that the shared-global functions have separate, manual declarations. We could do something clever with the macros (e.g., expand it to a declaration in some places, and a declaration _and_ a definition in path.c). But there aren't that many, and it's probably better to stay away from too-magical macros. Likewise, if we abandon the C preprocessor in favor of generating these with a script, we could get much fancier. E.g., normalizing "FOO/BAR-BAZ" into "git_path_foo_bar_baz". But the small amount of saved typing is probably not worth the resulting confusion to readers who want to grep for the function's definition. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
7 years ago
static GIT_PATH_FUNC(rebase_path, "rebase-merge")
/*
* The file containing rebase commands, comments, and empty lines.
* This file is created by "git rebase -i" then edited by the user. As
* the lines are processed, they are removed from the front of this
* file and written to the tail of 'done'.
*/
GIT_PATH_FUNC(rebase_path_todo, "rebase-merge/git-rebase-todo")
GIT_PATH_FUNC(rebase_path_todo_backup, "rebase-merge/git-rebase-todo.backup")
rebase-interactive: warn if commit is dropped with `rebase --edit-todo' When set to "warn" or "error", `rebase.missingCommitsCheck' would make `rebase -i' warn if the user removed commits from the todo list to prevent mistakes. Unfortunately, `rebase --edit-todo' and `rebase --continue' don't take it into account. This adds the ability for `rebase --edit-todo' and `rebase --continue' to check if commits were dropped by the user. As both edit_todo_list() and complete_action() parse the todo list and check for dropped commits, the code doing so in the latter is removed to reduce duplication. `edit_todo_list_advice' is removed from sequencer.c as it is no longer used there. This changes when a backup of the todo list is made. Until now, it was saved only once, before the initial edit. Now, it is also made if the original todo list has no errors or no dropped commits. Thus, the backup should be error-free. Without this, sequencer_continue() (`rebase --continue') could only compare the current todo list against the original, unedited list. Before this change, this file was only used by edit_todo_list() and `rebase -p' to create the backup before the initial edit, and check_todo_list_from_file(), only used by `rebase -p' to check for dropped commits after its own initial edit. If the edited list has an error, a file, `dropped', is created to report the issue. Otherwise, it is deleted. Usually, the edited list is compared against the list before editing, but if this file exists, it will be compared to the backup. Also, if the file exists, sequencer_continue() checks the list for dropped commits. If the check was performed every time, it would fail when resuming a rebase after resolving a conflict, as the backup will contain commits that were picked, but they will not be in the new list. It's safe to ignore this check if `dropped' does not exist, because that means that no errors were found at the last edition, so any missing commits here have already been picked. Five tests are added to t3404. The tests for `rebase.missingCommitsCheck = warn' and `rebase.missingCommitsCheck = error' have a similar structure. First, we start a rebase with an incorrect command on the first line. Then, we edit the todo list, removing the first and the last lines. This demonstrates that `--edit-todo' notices dropped commits, but not when the command is incorrect. Then, we restore the original todo list, and edit it to remove the last line. This demonstrates that if we add a commit after the initial edit, then remove it, `--edit-todo' will notice that it has been dropped. Then, the actual rebase takes place. In the third test, it is also checked that `--continue' will refuse to resume the rebase if commits were dropped. The fourth test checks that no errors are raised when resuming a rebase after resolving a conflict, the fifth checks that no errors are raised when editing the todo list after pausing the rebase. Signed-off-by: Alban Gruin <alban.gruin@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years ago
GIT_PATH_FUNC(rebase_path_dropped, "rebase-merge/dropped")
/*
* The rebase command lines that have already been processed. A line
* is moved here when it is first handled, before any associated user
* actions.
*/
static GIT_PATH_FUNC(rebase_path_done, "rebase-merge/done")
/*
* The file to keep track of how many commands were already processed (e.g.
* for the prompt).
*/
static GIT_PATH_FUNC(rebase_path_msgnum, "rebase-merge/msgnum")
/*
* The file to keep track of how many commands are to be processed in total
* (e.g. for the prompt).
*/
static GIT_PATH_FUNC(rebase_path_msgtotal, "rebase-merge/end")
sequencer (rebase -i): add support for the 'fixup' and 'squash' commands This is a huge patch, and at the same time a huge step forward to execute the performance-critical parts of the interactive rebase in a builtin command. Since 'fixup' and 'squash' are not only similar, but also need to know about each other (we want to reduce a series of fixups/squashes into a single, final commit message edit, from the user's point of view), we really have to implement them both at the same time. Most of the actual work is done by the existing code path that already handles the "pick" and the "edit" commands; We added support for other features (e.g. to amend the commit message) in the patches leading up to this one, yet there are still quite a few bits in this patch that simply would not make sense as individual patches (such as: determining whether there was anything to "fix up" in the "todo" script, etc). In theory, it would be possible to reuse the fast-forward code path also for the fixup and the squash code paths, but in practice this would make the code less readable. The end result cannot be fast-forwarded anyway, therefore let's just extend the cherry-picking code path for now. Since the sequencer parses the entire `git-rebase-todo` script in one go, fixup or squash commands without a preceding pick can be reported early (in git-rebase--interactive, we could only report such errors just before executing the fixup/squash). Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
6 years ago
/*
* The commit message that is planned to be used for any changes that
* need to be committed following a user interaction.
*/
static GIT_PATH_FUNC(rebase_path_message, "rebase-merge/message")
/*
* The file into which is accumulated the suggested commit message for
* squash/fixup commands. When the first of a series of squash/fixups
* is seen, the file is created and the commit message from the
* previous commit and from the first squash/fixup commit are written
* to it. The commit message for each subsequent squash/fixup commit
* is appended to the file as it is processed.
*/
static GIT_PATH_FUNC(rebase_path_squash_msg, "rebase-merge/message-squash")
/*
* If the current series of squash/fixups has not yet included a squash
* command, then this file exists and holds the commit message of the
* original "pick" commit. (If the series ends without a "squash"
* command, then this can be used as the commit message of the combined
* commit without opening the editor.)
*/
static GIT_PATH_FUNC(rebase_path_fixup_msg, "rebase-merge/message-fixup")
/*
* This file contains the list fixup/squash commands that have been
* accumulated into message-fixup or message-squash so far.
*/
static GIT_PATH_FUNC(rebase_path_current_fixups, "rebase-merge/current-fixups")
/*
* A script to set the GIT_AUTHOR_NAME, GIT_AUTHOR_EMAIL, and
* GIT_AUTHOR_DATE that will be used for the commit that is currently
* being rebased.
*/
static GIT_PATH_FUNC(rebase_path_author_script, "rebase-merge/author-script")
/*
* When an "edit" rebase command is being processed, the SHA1 of the
* commit to be edited is recorded in this file. When "git rebase
* --continue" is executed, if there are any staged changes then they
* will be amended to the HEAD commit, but only provided the HEAD
* commit is still the commit to be edited. When any other rebase
* command is processed, this file is deleted.
*/
static GIT_PATH_FUNC(rebase_path_amend, "rebase-merge/amend")
/*
* When we stop at a given patch via the "edit" command, this file contains
* the commit object name of the corresponding patch.
*/
static GIT_PATH_FUNC(rebase_path_stopped_sha, "rebase-merge/stopped-sha")
/*
* For the post-rewrite hook, we make a list of rewritten commits and
* their new sha1s. The rewritten-pending list keeps the sha1s of
* commits that have been processed, but not committed yet,
* e.g. because they are waiting for a 'squash' command.
*/
static GIT_PATH_FUNC(rebase_path_rewritten_list, "rebase-merge/rewritten-list")
static GIT_PATH_FUNC(rebase_path_rewritten_pending,
"rebase-merge/rewritten-pending")
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
/*
* The path of the file containing the OID of the "squash onto" commit, i.e.
* the dummy commit used for `reset [new root]`.
*/
static GIT_PATH_FUNC(rebase_path_squash_onto, "rebase-merge/squash-onto")
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
/*
* The path of the file listing refs that need to be deleted after the rebase
* finishes. This is used by the `label` command to record the need for cleanup.
*/
static GIT_PATH_FUNC(rebase_path_refs_to_delete, "rebase-merge/refs-to-delete")
2 months ago
/*
* The update-refs file stores a list of refs that will be updated at the end
* of the rebase sequence. The 'update-ref <ref>' commands in the todo file
* update the OIDs for the refs in this file, but the refs are not updated
* until the end of the rebase sequence.
*
* rebase_path_update_refs() returns the path to this file for a given
* worktree directory. For the current worktree, pass the_repository->gitdir.
*/
static char *rebase_path_update_refs(const char *wt_git_dir)
{
return xstrfmt("%s/rebase-merge/update-refs", wt_git_dir);
}
/*
* The following files are written by git-rebase just after parsing the
* command-line.
*/
static GIT_PATH_FUNC(rebase_path_gpg_sign_opt, "rebase-merge/gpg_sign_opt")
static GIT_PATH_FUNC(rebase_path_cdate_is_adate, "rebase-merge/cdate_is_adate")
static GIT_PATH_FUNC(rebase_path_ignore_date, "rebase-merge/ignore_date")
static GIT_PATH_FUNC(rebase_path_orig_head, "rebase-merge/orig-head")
static GIT_PATH_FUNC(rebase_path_verbose, "rebase-merge/verbose")
static GIT_PATH_FUNC(rebase_path_quiet, "rebase-merge/quiet")
static GIT_PATH_FUNC(rebase_path_signoff, "rebase-merge/signoff")
static GIT_PATH_FUNC(rebase_path_head_name, "rebase-merge/head-name")
static GIT_PATH_FUNC(rebase_path_onto, "rebase-merge/onto")
static GIT_PATH_FUNC(rebase_path_autostash, "rebase-merge/autostash")
static GIT_PATH_FUNC(rebase_path_strategy, "rebase-merge/strategy")
static GIT_PATH_FUNC(rebase_path_strategy_opts, "rebase-merge/strategy_opts")
static GIT_PATH_FUNC(rebase_path_allow_rerere_autoupdate, "rebase-merge/allow_rerere_autoupdate")
static GIT_PATH_FUNC(rebase_path_reschedule_failed_exec, "rebase-merge/reschedule-failed-exec")
rebase: don't override --no-reschedule-failed-exec with config Fix a bug in how --no-reschedule-failed-exec interacts with rebase.rescheduleFailedExec=true being set in the config. Before this change the --no-reschedule-failed-exec config option would be overridden by the config. This bug happened because of the particulars of how "rebase" works v.s. most other git commands when it comes to parsing options and config: When we read the config and parse the CLI options we correctly prefer the --no-reschedule-failed-exec option over rebase.rescheduleFailedExec=true in the config. So far so good. However the --reschedule-failed-exec option doesn't take effect when the rebase starts (we'd just create a ".git/rebase-merge/reschedule-failed-exec" file if it was true). It only takes effect when the exec command fails, at which point we'll reschedule the failed "exec" command. Since we only wrote out the positive ".git/rebase-merge/reschedule-failed-exec" under --reschedule-failed-exec, but nothing with --no-reschedule-failed-exec we'll forget that we asked not to reschedule failed "exec", and would happily re-read the config and see that rebase.rescheduleFailedExec=true is set. So the config will effectively override the user having explicitly disabled the option on the command-line. Even more confusingly: Since rebase accepts different options based on its state there wasn't even a way to get around this with "rebase --continue --no-reschedule-failed-exec" (but you could of course set the config with "rebase -c ..."). I think the least bad way out of this is to declare that for such options and config whatever we decide at the beginning of the rebase goes. So we'll now always create either a "reschedule-failed-exec" or a "no-reschedule-failed-exec file at the start, not just the former if we decided we wanted the feature. With this new worldview you can no longer change the setting once a rebase has started except by manually removing the state files discussed above. I think making it work like that is the the least confusing thing we can do. In the future we might want to learn to change the setting in the middle by combining "--edit-todo" with "--[no-]reschedule-failed-exec", we currently don't support combining those options, or any other way to change the state in the middle of the rebase short of manually editing the files in ".git/rebase-merge/*". The bug being fixed here originally came about because of a combination of the behavior of the code added in d421afa0c66 (rebase: introduce --reschedule-failed-exec, 2018-12-10) and the addition of the config variable in 969de3ff0e0 (rebase: add a config option to default to --reschedule-failed-exec, 2018-12-10). Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
1 year ago
static GIT_PATH_FUNC(rebase_path_no_reschedule_failed_exec, "rebase-merge/no-reschedule-failed-exec")
rebase (interactive-backend): fix handling of commits that become empty As established in the previous commit and commit b00bf1c9a8dd (git-rebase: make --allow-empty-message the default, 2018-06-27), the behavior for rebase with different backends in various edge or corner cases is often more happenstance than design. This commit addresses another such corner case: commits which "become empty". A careful reader may note that there are two types of commits which would become empty due to a rebase: * [clean cherry-pick] Commits which are clean cherry-picks of upstream commits, as determined by `git log --cherry-mark ...`. Re-applying these commits would result in an empty set of changes and a duplicative commit message; i.e. these are commits that have "already been applied" upstream. * [become empty] Commits which are not empty to start, are not clean cherry-picks of upstream commits, but which still become empty after being rebased. This happens e.g. when a commit has changes which are a strict subset of the changes in an upstream commit, or when the changes of a commit can be found spread across or among several upstream commits. Clearly, in both cases the changes in the commit in question are found upstream already, but the commit message may not be in the latter case. When cherry-mark can determine a commit is already upstream, then because of how cherry-mark works this means the upstream commit message was about the *exact* same set of changes. Thus, the commit messages can be assumed to be fully interchangeable (and are in fact likely to be completely identical). As such, the clean cherry-pick case represents a case when there is no information to be gained by keeping the extra commit around. All rebase types have always dropped these commits, and no one to my knowledge has ever requested that we do otherwise. For many of the become empty cases (and likely even most), we will also be able to drop the commit without loss of information -- but this isn't quite always the case. Since these commits represent cases that were not clean cherry-picks, there is no upstream commit message explaining the same set of changes. Projects with good commit message hygiene will likely have the explanation from our commit message contained within or spread among the relevant upstream commits, but not all projects run that way. As such, the commit message of the commit being rebased may have reasoning that suggests additional changes that should be made to adapt to the new base, or it may have information that someone wants to add as a note to another commit, or perhaps someone even wants to create an empty commit with the commit message as-is. Junio commented on the "become-empty" types of commits as follows[1]: WRT a change that ends up being empty (as opposed to a change that is empty from the beginning), I'd think that the current behaviour is desireable one. "am" based rebase is solely to transplant an existing history and want to stop much less than "interactive" one whose purpose is to polish a series before making it publishable, and asking for confirmation ("this has become empty--do you want to drop it?") is more appropriate from the workflow point of view. [1] https://lore.kernel.org/git/xmqqfu1fswdh.fsf@gitster-ct.c.googlers.com/ I would simply add that his arguments for "am"-based rebases actually apply to all non-explicitly-interactive rebases. Also, since we are stating that different cases should have different defaults, it may be worth providing a flag to allow users to select which behavior they want for these commits. Introduce a new command line flag for selecting the desired behavior: --empty={drop,keep,ask} with the definitions: drop: drop commits which become empty keep: keep commits which become empty ask: provide the user a chance to interact and pick what to do with commits which become empty on a case-by-case basis In line with Junio's suggestion, if the --empty flag is not specified, pick defaults as follows: explicitly interactive: ask otherwise: drop Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
3 years ago
static GIT_PATH_FUNC(rebase_path_drop_redundant_commits, "rebase-merge/drop_redundant_commits")
static GIT_PATH_FUNC(rebase_path_keep_redundant_commits, "rebase-merge/keep_redundant_commits")
2 months ago
/**
* A 'struct update_refs_record' represents a value in the update-refs
* list. We use a string_list to map refs to these (before, after) pairs.
*/
struct update_ref_record {
struct object_id before;
struct object_id after;
};
static struct update_ref_record *init_update_ref_record(const char *ref)
{
struct update_ref_record *rec;
CALLOC_ARRAY(rec, 1);
oidcpy(&rec->before, null_oid());
oidcpy(&rec->after, null_oid());
/* This may fail, but that's fine, we will keep the null OID. */
read_ref(ref, &rec->before);
return rec;
}
static int git_sequencer_config(const char *k, const char *v, void *cb)
{
struct replay_opts *opts = cb;
int status;
if (!strcmp(k, "commit.cleanup")) {
const char *s;
status = git_config_string(&s, k, v);
if (status)
return status;
if (!strcmp(s, "verbatim")) {
opts->default_msg_cleanup = COMMIT_MSG_CLEANUP_NONE;
opts->explicit_cleanup = 1;
} else if (!strcmp(s, "whitespace")) {
opts->default_msg_cleanup = COMMIT_MSG_CLEANUP_SPACE;
opts->explicit_cleanup = 1;
} else if (!strcmp(s, "strip")) {
opts->default_msg_cleanup = COMMIT_MSG_CLEANUP_ALL;
opts->explicit_cleanup = 1;
} else if (!strcmp(s, "scissors")) {
opts->default_msg_cleanup = COMMIT_MSG_CLEANUP_SCISSORS;
opts->explicit_cleanup = 1;
} else {
warning(_("invalid commit message cleanup mode '%s'"),
s);
}
free((char *)s);
return status;
}
if (!strcmp(k, "commit.gpgsign")) {
opts->gpg_sign = git_config_bool(k, v) ? xstrdup("") : NULL;
return 0;
}
if (!opts->default_strategy && !strcmp(k, "pull.twohead")) {
int ret = git_config_string((const char**)&opts->default_strategy, k, v);
if (ret == 0) {
/*
* pull.twohead is allowed to be multi-valued; we only
* care about the first value.
*/
char *tmp = strchr(opts->default_strategy, ' ');
if (tmp)
*tmp = '\0';
}
return ret;
}
if (opts->action == REPLAY_REVERT && !strcmp(k, "revert.reference"))
revert: optionally refer to commit in the "reference" format A typical "git revert" commit uses the full title of the original commit in its title, and starts its body of the message with: This reverts commit 8fa7f667cf61386257c00d6e954855cc3215ae91. This does not encourage the best practice of describing not just "what" (i.e. "Revert X" on the title says what we did) but "why" (i.e. and it does not say why X was undesirable). We can instead phrase this first line of the body to be more like This reverts commit 8fa7f667 (do this and that, 2022-04-25) so that the title does not have to be Revert "do this and that" We can instead use the title to describe "why" we are reverting the original commit. Introduce the "--reference" option to "git revert", and also the revert.reference configuration variable, which defaults to false, to tweak the title and the first line of the draft commit message for when creating a "revert" commit. When this option is in use, the first line of the pre-filled editor buffer becomes a comment line that tells the user to say _why_. If the user exits the editor without touching this line by mistake, what we prepare to become the first line of the body, i.e. "This reverts commit 8fa7f667 (do this and that, 2022-04-25)", ends up to be the title of the resulting commit. This behaviour is designed to help such a user to identify such a revert in "git log --oneline" easily so that it can be further reworded with "git rebase -i" later. Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 months ago
opts->commit_use_reference = git_config_bool(k, v);
status = git_gpg_config(k, v, NULL);
if (status)
return status;
return git_diff_basic_config(k, v, NULL);
}
void sequencer_init_config(struct replay_opts *opts)
{
opts->default_msg_cleanup = COMMIT_MSG_CLEANUP_NONE;
git_config(git_sequencer_config, opts);
}
static inline int is_rebase_i(const struct replay_opts *opts)
{
return opts->action == REPLAY_INTERACTIVE_REBASE;
}
static const char *get_dir(const struct replay_opts *opts)
{
if (is_rebase_i(opts))
return rebase_path();
return git_path_seq_dir();
}
static const char *get_todo_path(const struct replay_opts *opts)
{
if (is_rebase_i(opts))
return rebase_path_todo();
return git_path_todo_file();
}
/*
* Returns 0 for non-conforming footer
* Returns 1 for conforming footer
* Returns 2 when sob exists within conforming footer
* Returns 3 when sob exists within conforming footer as last entry
*/
static int has_conforming_footer(struct strbuf *sb, struct strbuf *sob,
size_t ignore_footer)
{
struct process_trailer_options opts = PROCESS_TRAILER_OPTIONS_INIT;
struct trailer_info info;
size_t i;
int found_sob = 0, found_sob_last = 0;
sequencer: handle ignore_footer when parsing trailers The append_signoff() function takes an "ignore_footer" argument, which specifies a number of bytes at the end of the message buffer which should not be considered (they cannot contain trailers, and the trailer is spliced in before them). But to find the existing trailers, it calls into has_conforming_trailer(). That function takes an ignore_footer parameter, but since 967dfd4d56 (sequencer: use trailer's trailer layout, 2016-11-02) the parameter is completely ignored. The trailer interface we're using takes a single string, with no option to tell it to use part of the string. However, since we have a mutable strbuf, we can work around this by simply overwriting (and later restoring) the boundary with a NUL. I'm not sure if this can actually trigger a bug in practice. It's easy to get a non-zero ignore_footer by doing something like this: git commit -F - --cleanup=verbatim <<-EOF subject body Signed-off-by: me # this looks like a comment, but is actually in the # message! That makes the earlier s-o-b fake. EOF git commit --amend -s There git-commit calls ignore_non_trailer() to count up the "#" cruft, which becomes the ignore_footer header. But it works even without this patch! That's because the trailer code _also_ calls ignore_non_trailer() and skips the cruft, too. So it happens to work because the only callers with a non-zero ignore_footer are using the exact same function that the trailer parser uses internally. And that seems true for all of the current callers, but there's nothing guaranteeing it. We're better off only feeding the correct buffer to the trailer code in the first place. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years ago
char saved_char;
opts.no_divider = 1;
sequencer: handle ignore_footer when parsing trailers The append_signoff() function takes an "ignore_footer" argument, which specifies a number of bytes at the end of the message buffer which should not be considered (they cannot contain trailers, and the trailer is spliced in before them). But to find the existing trailers, it calls into has_conforming_trailer(). That function takes an ignore_footer parameter, but since 967dfd4d56 (sequencer: use trailer's trailer layout, 2016-11-02) the parameter is completely ignored. The trailer interface we're using takes a single string, with no option to tell it to use part of the string. However, since we have a mutable strbuf, we can work around this by simply overwriting (and later restoring) the boundary with a NUL. I'm not sure if this can actually trigger a bug in practice. It's easy to get a non-zero ignore_footer by doing something like this: git commit -F - --cleanup=verbatim <<-EOF subject body Signed-off-by: me # this looks like a comment, but is actually in the # message! That makes the earlier s-o-b fake. EOF git commit --amend -s There git-commit calls ignore_non_trailer() to count up the "#" cruft, which becomes the ignore_footer header. But it works even without this patch! That's because the trailer code _also_ calls ignore_non_trailer() and skips the cruft, too. So it happens to work because the only callers with a non-zero ignore_footer are using the exact same function that the trailer parser uses internally. And that seems true for all of the current callers, but there's nothing guaranteeing it. We're better off only feeding the correct buffer to the trailer code in the first place. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years ago
if (ignore_footer) {
saved_char = sb->buf[sb->len - ignore_footer];
sb->buf[sb->len - ignore_footer] = '\0';
}
trailer_info_get(&info, sb->buf, &opts);
sequencer: handle ignore_footer when parsing trailers The append_signoff() function takes an "ignore_footer" argument, which specifies a number of bytes at the end of the message buffer which should not be considered (they cannot contain trailers, and the trailer is spliced in before them). But to find the existing trailers, it calls into has_conforming_trailer(). That function takes an ignore_footer parameter, but since 967dfd4d56 (sequencer: use trailer's trailer layout, 2016-11-02) the parameter is completely ignored. The trailer interface we're using takes a single string, with no option to tell it to use part of the string. However, since we have a mutable strbuf, we can work around this by simply overwriting (and later restoring) the boundary with a NUL. I'm not sure if this can actually trigger a bug in practice. It's easy to get a non-zero ignore_footer by doing something like this: git commit -F - --cleanup=verbatim <<-EOF subject body Signed-off-by: me # this looks like a comment, but is actually in the # message! That makes the earlier s-o-b fake. EOF git commit --amend -s There git-commit calls ignore_non_trailer() to count up the "#" cruft, which becomes the ignore_footer header. But it works even without this patch! That's because the trailer code _also_ calls ignore_non_trailer() and skips the cruft, too. So it happens to work because the only callers with a non-zero ignore_footer are using the exact same function that the trailer parser uses internally. And that seems true for all of the current callers, but there's nothing guaranteeing it. We're better off only feeding the correct buffer to the trailer code in the first place. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years ago
if (ignore_footer)
sb->buf[sb->len - ignore_footer] = saved_char;
if (info.trailer_start == info.trailer_end)
return 0;
for (i = 0; i < info.trailer_nr; i++)
if (sob && !strncmp(info.trailers[i], sob->buf, sob->len)) {
found_sob = 1;
if (i == info.trailer_nr - 1)
found_sob_last = 1;
}
trailer_info_release(&info);
if (found_sob_last)
return 3;
if (found_sob)
return 2;
return 1;
}
static const char *gpg_sign_opt_quoted(struct replay_opts *opts)
{
static struct strbuf buf = STRBUF_INIT;
strbuf_reset(&buf);
if (opts->gpg_sign)
sq_quotef(&buf, "-S%s", opts->gpg_sign);
return buf.buf;
}
int sequencer_remove_state(struct replay_opts *opts)
{
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
struct strbuf buf = STRBUF_INIT;
int i, ret = 0;
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
if (is_rebase_i(opts) &&
strbuf_read_file(&buf, rebase_path_refs_to_delete(), 0) > 0) {
char *p = buf.buf;
while (*p) {
char *eol = strchr(p, '\n');
if (eol)
*eol = '\0';
if (delete_ref("(rebase) cleanup", p, NULL, 0) < 0) {
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
warning(_("could not delete '%s'"), p);
ret = -1;
}
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
if (!eol)
break;
p = eol + 1;
}
}
free(opts->gpg_sign);
free(opts->default_strategy);
free(opts->strategy);
for (i = 0; i < opts->xopts_nr; i++)
free(opts->xopts[i]);
free(opts->xopts);
strbuf_release(&opts->current_fixups);
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
strbuf_reset(&buf);
strbuf_addstr(&buf, get_dir(opts));
if (remove_dir_recursively(&buf, 0))
ret = error(_("could not remove '%s'"), buf.buf);
sequencer: introduce new commands to reset the revision In the upcoming commits, we will teach the sequencer to rebase merges. This will be done in a very different way from the unfortunate design of `git rebase --preserve-merges` (which does not allow for reordering commits, or changing the branch topology). The main idea is to introduce new todo list commands, to support labeling the current revision with a given name, resetting the current revision to a previous state, and merging labeled revisions. This idea was developed in Git for Windows' Git garden shears (that are used to maintain Git for Windows' "thicket of branches" on top of upstream Git), and this patch is part of the effort to make it available to a wider audience, as well as to make the entire process more robust (by implementing it in a safe and portable language rather than a Unix shell script). This commit implements the commands to label, and to reset to, given revisions. The syntax is: label <name> reset <name> Internally, the `label <name>` command creates the ref `refs/rewritten/<name>`. This makes it possible to work with the labeled revisions interactively, or in a scripted fashion (e.g. via the todo list command `exec`). These temporary refs are removed upon sequencer_remove_state(), so that even a `git rebase --abort` cleans them up. We disallow '#' as label because that character will be used as separator in the upcoming `merge` command. Later in this patch series, we will mark the `refs/rewritten/` refs as worktree-local, to allow for interactive rebases to be run in parallel in worktrees linked to the same repository. As typos happen, a failed `label` or `reset` command will be rescheduled immediately. As the previous code to reschedule a command is embedded deeply in the pick/fixup/squash code path, we simply duplicate the few lines. This will allow us to extend the new code path easily for the upcoming `merge` command. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
4 years ago
strbuf_release(&buf);
return ret;
}
static const char *action_name(const struct replay_opts *opts)
{
switch (opts->action) {
case REPLAY_REVERT:
return N_("revert");
case REPLAY_PICK:
return N_("cherry-pick");
case REPLAY_INTERACTIVE_REBASE:
return N_("rebase");
}
die(_("unknown action: %d"), opts->action);
}
struct commit_message {
char *parent_label;
char *label;
char *subject;
const char *message;
};
static const char *short_commit_name(struct commit *commit)
{
return find_unique_abbrev(&commit->object.oid, DEFAULT_ABBREV);
}
static int get_message(struct commit *commit, struct commit_message *out)
{
const char *abbrev, *subject;
int subject_len;
out->message = logmsg_reencode(commit, NULL, get_commit_output_encoding());
abbrev = short_commit_name(commit);
subject_len = find_commit_subject(out->message, &subject);
out->subject = xmemdupz(subject, subject_len);
sequencer: avoid garbled merge machinery messages due to commit labels sequencer's get_message() exists to provide good labels on conflict hunks; see commits d68565402a ("revert: clarify label on conflict hunks", 2010-03-20) bf975d379d ("cherry-pick, revert: add a label for ancestor", 2010-03-20) 043a4492b3 ("sequencer: factor code out of revert builtin", 2012-01-11). for background on this function. These labels are of the form <commitID>... <commit summary> or parent of <commitID>... <commit summary> These labels are then passed as branch names to the merge machinery. However, these labels, as formatted, often also serve to confuse. For example, if we have a rename involved in a content merge, then it results in text such as the following: <<<<<<<< HEAD:foo.c int j; ======== int counter; >>>>>>>> b01dface... Removed unnecessary stuff:bar.c Or in various conflict messages, it can make it very difficult to read: CONFLICT (rename/delete): foo.c deleted in b01dface... Removed unnecessary stuff and renamed in HEAD. Version HEAD of foo.c left in tree. CONFLICT (file location): dir1/foo.c added in b01dface... Removed unnecessary stuff inside a directory that was renamed in HEAD, suggesting it should perhaps be moved to dir2/foo.c. Make a minor change to remove the ellipses and add parentheses around the commit summary; this makes all three examples much easier to read: <<<<<<<< HEAD:foo.c int j; ======== int counter; >>>>>>>> b01dface (Removed unnecessary stuff):bar.c CONFLICT (rename/delete): foo.c deleted in b01dface (Removed unnecessary stuff) and renamed in HEAD. Version HEAD of foo.c left in tree. CONFLICT (file location): dir1/foo.c added in b01dface (Removed unnecessary stuff) inside a directory that was renamed in HEAD, suggesting it should perhaps be moved to dir2/foo.c. Signed-off-by: Elijah Newren <newren@gmail.com> Reviewed-by: Taylor Blau <me@ttaylorr.com> Acked-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2 years ago
out->label = xstrfmt("%s (%s)", abbrev, out->subject);
out->parent_label = xstrfmt("parent of %s", out->label);
return 0;
}
static void free_message(struct commit *commit, struct commit_message *msg)
{