git/builtin/checkout.c

2054 lines
62 KiB
C
Raw Normal View History

#define USE_THE_INDEX_VARIABLE
#include "builtin.h"
#include "advice.h"
#include "branch.h"
#include "cache-tree.h"
#include "checkout.h"
#include "commit.h"
#include "config.h"
#include "diff.h"
#include "dir.h"
#include "environment.h"
#include "gettext.h"
#include "hex.h"
#include "hook.h"
merge-ll: rename from ll-merge A long term (but rather minor) pet-peeve of mine was the name ll-merge.[ch]. I thought it made it harder to realize what stuff was related to merging when I was working on the merge machinery and trying to improve it. Further, back in d1cbe1e6d8a ("hash-ll.h: split out of hash.h to remove dependency on repository.h", 2023-04-22), we have split the portions of hash.h that do not depend upon repository.h into a "hash-ll.h" (due to the recommendation to use "ll" for "low-level" in its name[1], but which I used as a suffix precisely because of my distaste for "ll-merge"). When we discussed adding additional "*-ll.h" files, a request was made that we use "ll" consistently as either a prefix or a suffix. Since it is already in use as both a prefix and a suffix, the only way to do so is to rename some files. Besides my distaste for the ll-merge.[ch] name, let me also note that the files ll-fsmonitor.h, ll-hash.h, ll-merge.h, ll-object-store.h, ll-read-cache.h would have essentially nothing to do with each other and make no sense to group. But giving them the common "ll-" prefix would group them. Using "-ll" as a suffix thus seems just much more logical to me. Rename ll-merge.[ch] to merge-ll.[ch] to achieve this consistency, and to ensure we get a more logical grouping of files. [1] https://lore.kernel.org/git/kl6lsfcu1g8w.fsf@chooglen-macbookpro.roam.corp.google.com/ Signed-off-by: Elijah Newren <newren@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-05-16 06:34:04 +00:00
#include "merge-ll.h"
#include "lockfile.h"
#include "mem-pool.h"
#include "merge-recursive.h"
#include "object-name.h"
#include "object-store-ll.h"
#include "parse-options.h"
#include "path.h"
#include "preload-index.h"
#include "read-cache.h"
#include "refs.h"
#include "remote.h"
#include "resolve-undo.h"
#include "revision.h"
#include "setup.h"
#include "submodule.h"
#include "symlinks.h"
#include "trace2.h"
#include "tree.h"
#include "tree-walk.h"
#include "unpack-trees.h"
#include "wt-status.h"
#include "xdiff-interface.h"
#include "entry.h"
#include "parallel-checkout.h"
#include "add-interactive.h"
static const char * const checkout_usage[] = {
N_("git checkout [<options>] <branch>"),
N_("git checkout [<options>] [<branch>] -- <file>..."),
NULL,
};
static const char * const switch_branch_usage[] = {
N_("git switch [<options>] [<branch>]"),
NULL,
};
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
static const char * const restore_usage[] = {
N_("git restore [<options>] [--source=<branch>] <file>..."),
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
NULL,
};
struct checkout_opts {
int patch_mode;
int quiet;
int merge;
int force;
int force_detach;
int implicit_detach;
int writeout_stage;
int overwrite_ignore;
int ignore_skipworktree;
int ignore_other_worktrees;
int show_progress;
int count_checkout_paths;
int overlay_mode;
int dwim_new_local_branch;
int discard_changes;
int accept_ref;
int accept_pathspec;
int switch_branch_doing_nothing_is_ok;
int only_merge_on_switching_branches;
int can_switch_when_in_progress;
int orphan_from_empty_tree;
int empty_pathspec_ok;
int checkout_index;
int checkout_worktree;
const char *ignore_unmerged_opt;
int ignore_unmerged;
int pathspec_file_nul;
parse-options: consistently allocate memory in fix_filename() When handling OPT_FILENAME(), we have to stick the "prefix" (if any) in front of the filename to make up for the fact that Git has chdir()'d to the top of the repository. We can do this with prefix_filename(), but there are a few special cases we handle ourselves. Unfortunately the memory allocation is inconsistent here; if we do make it to prefix_filename(), we'll allocate a string which the caller must free to avoid a leak. But if we hit our special cases, we'll return the string as-is, and a caller which tries to free it will crash. So there's no way to win. Let's consistently allocate, so that callers can do the right thing. There are now three cases to care about in the function (and hence a three-armed if/else): 1. we got a NULL input (and should leave it as NULL, though arguably this is the sign of a bug; let's keep the status quo for now and we can pick at that scab later) 2. we hit a special case that means we leave the name intact; we should duplicate the string. This includes our special "-" matching. Prior to this patch, it also included empty prefixes and absolute filenames. But we can observe that prefix_filename() already handles these, so we don't need to detect them. 3. everything else goes to prefix_filename() I've dropped the "const" from the "char **file" parameter to indicate that we're allocating, though in practice it's not really important. This is all being shuffled through a void pointer via opt->value before it hits code which ever looks at the string. And it's even a bit weird, because we are really taking _in_ a const string and using the same out-parameter for a non-const string. A better function signature would be: static char *fix_filename(const char *prefix, const char *file); but that would mean the caller dereferences the double-pointer (and the NULL check is currently handled inside this function). So I took the path of least-change here. Note that we have to fix several callers in this commit, too, or we'll break the leak-checking tests. These are "new" leaks in the sense that they are now triggered by the test suite, but these spots have always been leaky when Git is run in a subdirectory of the repository. I fixed all of the cases that trigger with GIT_TEST_PASSING_SANITIZE_LEAK. There may be others in scripts that have other leaks, but we can fix them later along with those other leaks (and again, you _couldn't_ fix them before this patch, so this is the necessary first step). Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-03-04 10:31:22 +00:00
char *pathspec_from_file;
const char *new_branch;
const char *new_branch_force;
const char *new_orphan_branch;
int new_branch_log;
enum branch_track track;
struct diff_options diff_options;
int conflict_style;
int branch_exists;
const char *prefix;
struct pathspec pathspec;
const char *from_treeish;
struct tree *source_tree;
};
#define CHECKOUT_OPTS_INIT { .conflict_style = -1, .merge = -1 }
struct branch_info {
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
char *name; /* The short name used */
char *path; /* The full name of a real branch */
struct commit *commit; /* The named commit */
char *refname; /* The full name of the ref being checked out. */
struct object_id oid; /* The object ID of the commit being checked out. */
/*
* if not null the branch is detached because it's already
* checked out in this checkout
*/
char *checkout;
};
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
static void branch_info_release(struct branch_info *info)
{
free(info->name);
free(info->path);
free(info->refname);
free(info->checkout);
}
static int post_checkout_hook(struct commit *old_commit, struct commit *new_commit,
int changed)
{
return run_hooks_l("post-checkout",
oid_to_hex(old_commit ? &old_commit->object.oid : null_oid()),
oid_to_hex(new_commit ? &new_commit->object.oid : null_oid()),
changed ? "1" : "0", NULL);
/* "new_commit" can be NULL when checking out from the index before
a commit exists. */
}
static int update_some(const struct object_id *oid, struct strbuf *base,
const char *pathname, unsigned mode, void *context UNUSED)
{
int len;
struct cache_entry *ce;
int pos;
if (S_ISDIR(mode))
return READ_TREE_RECURSIVE;
len = base->len + strlen(pathname);
block alloc: add lifecycle APIs for cache_entry structs It has been observed that the time spent loading an index with a large number of entries is partly dominated by malloc() calls. This change is in preparation for using memory pools to reduce the number of malloc() calls made to allocate cahce entries when loading an index. Add an API to allocate and discard cache entries, abstracting the details of managing the memory backing the cache entries. This commit does actually change how memory is managed - this will be done in a later commit in the series. This change makes the distinction between cache entries that are associated with an index and cache entries that are not associated with an index. A main use of cache entries is with an index, and we can optimize the memory management around this. We still have other cases where a cache entry is not persisted with an index, and so we need to handle the "transient" use case as well. To keep the congnitive overhead of managing the cache entries, there will only be a single discard function. This means there must be enough information kept with the cache entry so that we know how to discard them. A summary of the main functions in the API is: make_cache_entry: create cache entry for use in an index. Uses specified parameters to populate cache_entry fields. make_empty_cache_entry: Create an empty cache entry for use in an index. Returns cache entry with empty fields. make_transient_cache_entry: create cache entry that is not used in an index. Uses specified parameters to populate cache_entry fields. make_empty_transient_cache_entry: create cache entry that is not used in an index. Returns cache entry with empty fields. discard_cache_entry: A single function that knows how to discard a cache entry regardless of how it was allocated. Signed-off-by: Jameson Miller <jamill@microsoft.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-07-02 19:49:31 +00:00
ce = make_empty_cache_entry(&the_index, len);
oidcpy(&ce->oid, oid);
memcpy(ce->name, base->buf, base->len);
memcpy(ce->name + base->len, pathname, len - base->len);
ce->ce_flags = create_ce_flags(0) | CE_UPDATE;
ce->ce_namelen = len;
ce->ce_mode = create_ce_mode(mode);
/*
* If the entry is the same as the current index, we can leave the old
* entry in place. Whether it is UPTODATE or not, checkout_entry will
* do the right thing.
*/
pos = index_name_pos(&the_index, ce->name, ce->ce_namelen);
if (pos >= 0) {
struct cache_entry *old = the_index.cache[pos];
if (ce->ce_mode == old->ce_mode &&
!ce_intent_to_add(old) &&
oideq(&ce->oid, &old->oid)) {
old->ce_flags |= CE_UPDATE;
block alloc: add lifecycle APIs for cache_entry structs It has been observed that the time spent loading an index with a large number of entries is partly dominated by malloc() calls. This change is in preparation for using memory pools to reduce the number of malloc() calls made to allocate cahce entries when loading an index. Add an API to allocate and discard cache entries, abstracting the details of managing the memory backing the cache entries. This commit does actually change how memory is managed - this will be done in a later commit in the series. This change makes the distinction between cache entries that are associated with an index and cache entries that are not associated with an index. A main use of cache entries is with an index, and we can optimize the memory management around this. We still have other cases where a cache entry is not persisted with an index, and so we need to handle the "transient" use case as well. To keep the congnitive overhead of managing the cache entries, there will only be a single discard function. This means there must be enough information kept with the cache entry so that we know how to discard them. A summary of the main functions in the API is: make_cache_entry: create cache entry for use in an index. Uses specified parameters to populate cache_entry fields. make_empty_cache_entry: Create an empty cache entry for use in an index. Returns cache entry with empty fields. make_transient_cache_entry: create cache entry that is not used in an index. Uses specified parameters to populate cache_entry fields. make_empty_transient_cache_entry: create cache entry that is not used in an index. Returns cache entry with empty fields. discard_cache_entry: A single function that knows how to discard a cache entry regardless of how it was allocated. Signed-off-by: Jameson Miller <jamill@microsoft.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-07-02 19:49:31 +00:00
discard_cache_entry(ce);
return 0;
}
}
add_index_entry(&the_index, ce,
ADD_CACHE_OK_TO_ADD | ADD_CACHE_OK_TO_REPLACE);
return 0;
}
static int read_tree_some(struct tree *tree, const struct pathspec *pathspec)
{
read_tree(the_repository, tree,
pathspec, update_some, NULL);
/* update the index with the given tree's info
* for all args, expanding wildcards, and exit
* with any non-zero return code.
*/
return 0;
}
Convert "struct cache_entry *" to "const ..." wherever possible I attempted to make index_state->cache[] a "const struct cache_entry **" to find out how existing entries in index are modified and where. The question I have is what do we do if we really need to keep track of on-disk changes in the index. The result is - diff-lib.c: setting CE_UPTODATE - name-hash.c: setting CE_HASHED - preload-index.c, read-cache.c, unpack-trees.c and builtin/update-index: obvious - entry.c: write_entry() may refresh the checked out entry via fill_stat_cache_info(). This causes "non-const struct cache_entry *" in builtin/apply.c, builtin/checkout-index.c and builtin/checkout.c - builtin/ls-files.c: --with-tree changes stagemask and may set CE_UPDATE Of these, write_entry() and its call sites are probably most interesting because it modifies on-disk info. But this is stat info and can be retrieved via refresh, at least for porcelain commands. Other just uses ce_flags for local purposes. So, keeping track of "dirty" entries is just a matter of setting a flag in index modification functions exposed by read-cache.c. Except unpack-trees, the rest of the code base does not do anything funny behind read-cache's back. The actual patch is less valueable than the summary above. But if anyone wants to re-identify the above sites. Applying this patch, then this: diff --git a/cache.h b/cache.h index 430d021..1692891 100644 --- a/cache.h +++ b/cache.h @@ -267,7 +267,7 @@ static inline unsigned int canon_mode(unsigned int mode) #define cache_entry_size(len) (offsetof(struct cache_entry,name) + (len) + 1) struct index_state { - struct cache_entry **cache; + const struct cache_entry **cache; unsigned int version; unsigned int cache_nr, cache_alloc, cache_changed; struct string_list *resolve_undo; will help quickly identify them without bogus warnings. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-07-09 15:29:00 +00:00
static int skip_same_name(const struct cache_entry *ce, int pos)
{
while (++pos < the_index.cache_nr &&
!strcmp(the_index.cache[pos]->name, ce->name))
; /* skip */
return pos;
}
static int check_stage(int stage, const struct cache_entry *ce, int pos,
int overlay_mode)
{
while (pos < the_index.cache_nr &&
!strcmp(the_index.cache[pos]->name, ce->name)) {
if (ce_stage(the_index.cache[pos]) == stage)
return 0;
pos++;
}
if (!overlay_mode)
return 0;
if (stage == 2)
return error(_("path '%s' does not have our version"), ce->name);
else
return error(_("path '%s' does not have their version"), ce->name);
}
Convert "struct cache_entry *" to "const ..." wherever possible I attempted to make index_state->cache[] a "const struct cache_entry **" to find out how existing entries in index are modified and where. The question I have is what do we do if we really need to keep track of on-disk changes in the index. The result is - diff-lib.c: setting CE_UPTODATE - name-hash.c: setting CE_HASHED - preload-index.c, read-cache.c, unpack-trees.c and builtin/update-index: obvious - entry.c: write_entry() may refresh the checked out entry via fill_stat_cache_info(). This causes "non-const struct cache_entry *" in builtin/apply.c, builtin/checkout-index.c and builtin/checkout.c - builtin/ls-files.c: --with-tree changes stagemask and may set CE_UPDATE Of these, write_entry() and its call sites are probably most interesting because it modifies on-disk info. But this is stat info and can be retrieved via refresh, at least for porcelain commands. Other just uses ce_flags for local purposes. So, keeping track of "dirty" entries is just a matter of setting a flag in index modification functions exposed by read-cache.c. Except unpack-trees, the rest of the code base does not do anything funny behind read-cache's back. The actual patch is less valueable than the summary above. But if anyone wants to re-identify the above sites. Applying this patch, then this: diff --git a/cache.h b/cache.h index 430d021..1692891 100644 --- a/cache.h +++ b/cache.h @@ -267,7 +267,7 @@ static inline unsigned int canon_mode(unsigned int mode) #define cache_entry_size(len) (offsetof(struct cache_entry,name) + (len) + 1) struct index_state { - struct cache_entry **cache; + const struct cache_entry **cache; unsigned int version; unsigned int cache_nr, cache_alloc, cache_changed; struct string_list *resolve_undo; will help quickly identify them without bogus warnings. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-07-09 15:29:00 +00:00
static int check_stages(unsigned stages, const struct cache_entry *ce, int pos)
{
unsigned seen = 0;
const char *name = ce->name;
while (pos < the_index.cache_nr) {
ce = the_index.cache[pos];
if (strcmp(name, ce->name))
break;
seen |= (1 << ce_stage(ce));
pos++;
}
if ((stages & seen) != stages)
return error(_("path '%s' does not have all necessary versions"),
name);
return 0;
}
static int checkout_stage(int stage, const struct cache_entry *ce, int pos,
const struct checkout *state, int *nr_checkouts,
int overlay_mode)
{
while (pos < the_index.cache_nr &&
!strcmp(the_index.cache[pos]->name, ce->name)) {
if (ce_stage(the_index.cache[pos]) == stage)
return checkout_entry(the_index.cache[pos], state,
NULL, nr_checkouts);
pos++;
}
if (!overlay_mode) {
read-tree: add "--super-prefix" option, eliminate global The "--super-prefix" option to "git" was initially added in [1] for use with "ls-files"[2], and shortly thereafter "submodule--helper"[3] and "grep"[4]. It wasn't until [5] that "read-tree" made use of it. At the time [5] made sense, but since then we've made "ls-files" recurse in-process in [6], "grep" in [7], and finally "submodule--helper" in the preceding commits. Let's also remove it from "read-tree", which allows us to remove the option to "git" itself. We can do this because the only remaining user of it is the submodule API, which will now invoke "read-tree" with its new "--super-prefix" option. It will only do so when the "submodule_move_head()" function is called. That "submodule_move_head()" function was then only invoked by "read-tree" itself, but now rather than setting an environment variable to pass "--super-prefix" between cmd_read_tree() we: - Set a new "super_prefix" in "struct unpack_trees_options". The "super_prefixed()" function in "unpack-trees.c" added in [5] will now use this, rather than get_super_prefix() looking up the environment variable we set earlier in the same process. - Add the same field to the "struct checkout", which is only needed to ferry the "super_prefix" in the "struct unpack_trees_options" all the way down to the "entry.c" callers of "submodule_move_head()". Those calls which used the super prefix all originated in "cmd_read_tree()". The only other caller is the "unlink_entry()" caller in "builtin/checkout.c", which now passes a "NULL". 1. 74866d75793 (git: make super-prefix option, 2016-10-07) 2. e77aa336f11 (ls-files: optionally recurse into submodules, 2016-10-07) 3. 89c86265576 (submodule helper: support super prefix, 2016-12-08) 4. 0281e487fd9 (grep: optionally recurse into submodules, 2016-12-16) 5. 3d415425c7b (unpack-trees: support super-prefix option, 2017-01-17) 6. 188dce131fa (ls-files: use repository object, 2017-06-22) 7. f9ee2fcdfa0 (grep: recurse in-process using 'struct repository', 2017-08-02) Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-12-20 12:39:56 +00:00
unlink_entry(ce, NULL);
return 0;
}
if (stage == 2)
return error(_("path '%s' does not have our version"), ce->name);
else
return error(_("path '%s' does not have their version"), ce->name);
}
static int checkout_merged(int pos, const struct checkout *state,
int *nr_checkouts, struct mem_pool *ce_mem_pool,
int conflict_style)
{
struct cache_entry *ce = the_index.cache[pos];
const char *path = ce->name;
mmfile_t ancestor, ours, theirs;
enum ll_merge_result merge_status;
int status;
struct object_id oid;
mmbuffer_t result_buf;
struct object_id threeway[3];
unsigned mode = 0;
struct ll_merge_options ll_opts = LL_MERGE_OPTIONS_INIT;
int renormalize = 0;
memset(threeway, 0, sizeof(threeway));
while (pos < the_index.cache_nr) {
int stage;
stage = ce_stage(ce);
if (!stage || strcmp(path, ce->name))
break;
oidcpy(&threeway[stage - 1], &ce->oid);
if (stage == 2)
mode = create_ce_mode(ce->ce_mode);
pos++;
ce = the_index.cache[pos];
}
if (is_null_oid(&threeway[1]) || is_null_oid(&threeway[2]))
return error(_("path '%s' does not have necessary versions"), path);
read_mmblob(&ancestor, &threeway[0]);
read_mmblob(&ours, &threeway[1]);
read_mmblob(&theirs, &threeway[2]);
git_config_get_bool("merge.renormalize", &renormalize);
ll_opts.renormalize = renormalize;
ll_opts.conflict_style = conflict_style;
merge_status = ll_merge(&result_buf, path, &ancestor, "base",
&ours, "ours", &theirs, "theirs",
state->istate, &ll_opts);
free(ancestor.ptr);
free(ours.ptr);
free(theirs.ptr);
if (merge_status == LL_MERGE_BINARY_CONFLICT)
warning("Cannot merge binary files: %s (%s vs. %s)",
path, "ours", "theirs");
if (merge_status < 0 || !result_buf.ptr) {
free(result_buf.ptr);
return error(_("path '%s': cannot merge"), path);
}
/*
* NEEDSWORK:
* There is absolutely no reason to write this as a blob object
* and create a phony cache entry. This hack is primarily to get
* to the write_entry() machinery that massages the contents to
* work-tree format and writes out which only allows it for a
* cache entry. The code in write_entry() needs to be refactored
* to allow us to feed a <buffer, size, mode> instead of a cache
* entry. Such a refactoring would help merge_recursive as well
* (it also writes the merge result to the object database even
* when it may contain conflicts).
*/
if (write_object_file(result_buf.ptr, result_buf.size, OBJ_BLOB, &oid))
die(_("Unable to add merge result for '%s'"), path);
free(result_buf.ptr);
ce = make_transient_cache_entry(mode, &oid, path, 2, ce_mem_pool);
if (!ce)
die(_("make_cache_entry failed for path '%s'"), path);
status = checkout_entry(ce, state, NULL, nr_checkouts);
return status;
}
static void mark_ce_for_checkout_overlay(struct cache_entry *ce,
char *ps_matched,
const struct checkout_opts *opts)
{
ce->ce_flags &= ~CE_MATCHED;
if (!opts->ignore_skipworktree && ce_skip_worktree(ce))
return;
if (opts->source_tree && !(ce->ce_flags & CE_UPDATE))
/*
* "git checkout tree-ish -- path", but this entry
* is in the original index but is not in tree-ish
* or does not match the pathspec; it will not be
* checked out to the working tree. We will not do
* anything to this entry at all.
*/
return;
/*
* Either this entry came from the tree-ish we are
* checking the paths out of, or we are checking out
* of the index.
*
* If it comes from the tree-ish, we already know it
* matches the pathspec and could just stamp
* CE_MATCHED to it from update_some(). But we still
* need ps_matched and read_tree (and
* eventually tree_entry_interesting) cannot fill
* ps_matched yet. Once it can, we can avoid calling
* match_pathspec() for _all_ entries when
* opts->source_tree != NULL.
*/
if (ce_path_match(&the_index, ce, &opts->pathspec, ps_matched))
ce->ce_flags |= CE_MATCHED;
}
static void mark_ce_for_checkout_no_overlay(struct cache_entry *ce,
char *ps_matched,
const struct checkout_opts *opts)
{
ce->ce_flags &= ~CE_MATCHED;
if (!opts->ignore_skipworktree && ce_skip_worktree(ce))
return;
if (ce_path_match(&the_index, ce, &opts->pathspec, ps_matched)) {
ce->ce_flags |= CE_MATCHED;
if (opts->source_tree && !(ce->ce_flags & CE_UPDATE))
/*
* In overlay mode, but the path is not in
* tree-ish, which means we should remove it
* from the index and the working tree.
*/
ce->ce_flags |= CE_REMOVE | CE_WT_REMOVE;
}
}
static int checkout_worktree(const struct checkout_opts *opts,
const struct branch_info *info)
{
struct checkout state = CHECKOUT_INIT;
int nr_checkouts = 0, nr_unmerged = 0;
int errs = 0;
int pos;
int pc_workers, pc_threshold;
struct mem_pool ce_mem_pool;
state.force = 1;
state.refresh_cache = 1;
state.istate = &the_index;
mem_pool_init(&ce_mem_pool, 0);
get_parallel_checkout_configs(&pc_workers, &pc_threshold);
init_checkout_metadata(&state.meta, info->refname,
info->commit ? &info->commit->object.oid : &info->oid,
NULL);
enable_delayed_checkout(&state);
if (pc_workers > 1)
init_parallel_checkout();
for (pos = 0; pos < the_index.cache_nr; pos++) {
struct cache_entry *ce = the_index.cache[pos];
if (ce->ce_flags & CE_MATCHED) {
if (!ce_stage(ce)) {
errs |= checkout_entry(ce, &state,
NULL, &nr_checkouts);
continue;
}
if (opts->writeout_stage)
errs |= checkout_stage(opts->writeout_stage,
ce, pos,
&state,
&nr_checkouts, opts->overlay_mode);
else if (opts->merge)
errs |= checkout_merged(pos, &state,
&nr_unmerged,
&ce_mem_pool,
opts->conflict_style);
pos = skip_same_name(ce, pos) - 1;
}
}
if (pc_workers > 1)
errs |= run_parallel_checkout(&state, pc_workers, pc_threshold,
NULL, NULL);
mem_pool_discard(&ce_mem_pool, should_validate_cache_entries());
remove_marked_cache_entries(&the_index, 1);
remove_scheduled_dirs();
errs |= finish_delayed_checkout(&state, opts->show_progress);
if (opts->count_checkout_paths) {
if (nr_unmerged)
fprintf_ln(stderr, Q_("Recreated %d merge conflict",
"Recreated %d merge conflicts",
nr_unmerged),
nr_unmerged);
if (opts->source_tree)
fprintf_ln(stderr, Q_("Updated %d path from %s",
"Updated %d paths from %s",
nr_checkouts),
nr_checkouts,
repo_find_unique_abbrev(the_repository, &opts->source_tree->object.oid,
DEFAULT_ABBREV));
else if (!nr_unmerged || nr_checkouts)
fprintf_ln(stderr, Q_("Updated %d path from the index",
"Updated %d paths from the index",
nr_checkouts),
nr_checkouts);
}
return errs;
}
static int checkout_paths(const struct checkout_opts *opts,
const struct branch_info *new_branch_info)
{
int pos;
static char *ps_matched;
struct object_id rev;
struct commit *head;
int errs = 0;
struct lock_file lock_file = LOCK_INIT;
int checkout_index;
trace2_cmd_mode(opts->patch_mode ? "patch" : "path");
if (opts->track != BRANCH_TRACK_UNSPECIFIED)
die(_("'%s' cannot be used with updating paths"), "--track");
if (opts->new_branch_log)
die(_("'%s' cannot be used with updating paths"), "-l");
if (opts->ignore_unmerged && opts->patch_mode)
die(_("'%s' cannot be used with updating paths"),
opts->ignore_unmerged_opt);
if (opts->force_detach)
die(_("'%s' cannot be used with updating paths"), "--detach");
if (opts->merge && opts->patch_mode)
die(_("options '%s' and '%s' cannot be used together"), "--merge", "--patch");
if (opts->ignore_unmerged && opts->merge)
die(_("options '%s' and '%s' cannot be used together"),
opts->ignore_unmerged_opt, "-m");
if (opts->new_branch)
die(_("Cannot update paths and switch to branch '%s' at the same time."),
opts->new_branch);
if (!opts->checkout_worktree && !opts->checkout_index)
die(_("neither '%s' or '%s' is specified"),
"--staged", "--worktree");
if (!opts->checkout_worktree && !opts->from_treeish)
die(_("'%s' must be used when '%s' is not specified"),
"--worktree", "--source");
/*
* Reject --staged option to the restore command when combined with
* merge-related options. Use the accept_ref flag to distinguish it
* from the checkout command, which does not accept --staged anyway.
*
* `restore --ours|--theirs --worktree --staged` could mean resolving
* conflicted paths to one side in both the worktree and the index,
* but does not currently.
*
* `restore --merge|--conflict=<style>` already recreates conflicts
* in both the worktree and the index, so adding --staged would be
* meaningless.
*/
if (!opts->accept_ref && opts->checkout_index) {
if (opts->writeout_stage)
die(_("'%s' or '%s' cannot be used with %s"),
"--ours", "--theirs", "--staged");
if (opts->merge)
die(_("'%s' or '%s' cannot be used with %s"),
"--merge", "--conflict", "--staged");
}
/*
* recreating unmerged index entries and writing out data from
* unmerged index entries would make no sense when checking out
* of a tree-ish.
*/
if ((opts->merge || opts->writeout_stage) && opts->source_tree)
die(_("'%s', '%s', or '%s' cannot be used when checking out of a tree"),
"--merge", "--ours", "--theirs");
if (opts->patch_mode) {
enum add_p_mode patch_mode;
builtin/checkout: fix `git checkout -p HEAD...` bug Running `git checkout -p` with a merge-base rev results in an error: $ git checkout -p HEAD... usage: git diff-index [-m] [--cached] [<common-diff-options>] <tree-ish> [<path>...] common diff options: -z output diff-raw with lines terminated with NUL. -p output patch format. -u synonym for -p. --patch-with-raw output both a patch and the diff-raw format. --stat show diffstat instead of patch. --numstat show numeric diffstat instead of patch. --patch-with-stat output a patch and prepend its diffstat. --name-only show only names of changed files. --name-status show names and status of changed files. --full-index show full object name on index lines. --abbrev=<n> abbreviate object names in diff-tree header and diff-raw. -R swap input file pairs. -B detect complete rewrites. -M detect renames. -C detect copies. --find-copies-harder try unchanged files as candidate for copy detection. -l<n> limit rename attempts up to <n> paths. -O<file> reorder diffs according to the <file>. -S<string> find filepair whose only one side contains the string. --pickaxe-all show all files diff when -S is used and hit is found. -a --text treat all files as text. Cannot close git diff-index --cached --numstat --summary HEAD... -- () at <redacted>/libexec/git-core/git-add--interactive line 183. This happens because checkout passes the literal argument (in the example, `HEAD...`) to diff-index which does not recognise merge-base revs. Fix this by using the hex of the found commit instead of the given name. Note that "HEAD" is handled specially in run_add_interactive() so it's explicitly not changed. Signed-off-by: Denton Liu <liu.denton@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-10-07 07:56:15 +00:00
const char *rev = new_branch_info->name;
char rev_oid[GIT_MAX_HEXSZ + 1];
/*
* Since rev can be in the form of `<a>...<b>` (which is not
* recognized by diff-index), we will always replace the name
* with the hex of the commit (whether it's in `...` form or
* not) for the run_add_interactive() machinery to work
* properly. However, there is special logic for the HEAD case
* so we mustn't replace that. Also, when we were given a
* tree-object, new_branch_info->commit would be NULL, but we
* do not have to do any replacement, either.
builtin/checkout: fix `git checkout -p HEAD...` bug Running `git checkout -p` with a merge-base rev results in an error: $ git checkout -p HEAD... usage: git diff-index [-m] [--cached] [<common-diff-options>] <tree-ish> [<path>...] common diff options: -z output diff-raw with lines terminated with NUL. -p output patch format. -u synonym for -p. --patch-with-raw output both a patch and the diff-raw format. --stat show diffstat instead of patch. --numstat show numeric diffstat instead of patch. --patch-with-stat output a patch and prepend its diffstat. --name-only show only names of changed files. --name-status show names and status of changed files. --full-index show full object name on index lines. --abbrev=<n> abbreviate object names in diff-tree header and diff-raw. -R swap input file pairs. -B detect complete rewrites. -M detect renames. -C detect copies. --find-copies-harder try unchanged files as candidate for copy detection. -l<n> limit rename attempts up to <n> paths. -O<file> reorder diffs according to the <file>. -S<string> find filepair whose only one side contains the string. --pickaxe-all show all files diff when -S is used and hit is found. -a --text treat all files as text. Cannot close git diff-index --cached --numstat --summary HEAD... -- () at <redacted>/libexec/git-core/git-add--interactive line 183. This happens because checkout passes the literal argument (in the example, `HEAD...`) to diff-index which does not recognise merge-base revs. Fix this by using the hex of the found commit instead of the given name. Note that "HEAD" is handled specially in run_add_interactive() so it's explicitly not changed. Signed-off-by: Denton Liu <liu.denton@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-10-07 07:56:15 +00:00
*/
if (rev && new_branch_info->commit && strcmp(rev, "HEAD"))
builtin/checkout: fix `git checkout -p HEAD...` bug Running `git checkout -p` with a merge-base rev results in an error: $ git checkout -p HEAD... usage: git diff-index [-m] [--cached] [<common-diff-options>] <tree-ish> [<path>...] common diff options: -z output diff-raw with lines terminated with NUL. -p output patch format. -u synonym for -p. --patch-with-raw output both a patch and the diff-raw format. --stat show diffstat instead of patch. --numstat show numeric diffstat instead of patch. --patch-with-stat output a patch and prepend its diffstat. --name-only show only names of changed files. --name-status show names and status of changed files. --full-index show full object name on index lines. --abbrev=<n> abbreviate object names in diff-tree header and diff-raw. -R swap input file pairs. -B detect complete rewrites. -M detect renames. -C detect copies. --find-copies-harder try unchanged files as candidate for copy detection. -l<n> limit rename attempts up to <n> paths. -O<file> reorder diffs according to the <file>. -S<string> find filepair whose only one side contains the string. --pickaxe-all show all files diff when -S is used and hit is found. -a --text treat all files as text. Cannot close git diff-index --cached --numstat --summary HEAD... -- () at <redacted>/libexec/git-core/git-add--interactive line 183. This happens because checkout passes the literal argument (in the example, `HEAD...`) to diff-index which does not recognise merge-base revs. Fix this by using the hex of the found commit instead of the given name. Note that "HEAD" is handled specially in run_add_interactive() so it's explicitly not changed. Signed-off-by: Denton Liu <liu.denton@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-10-07 07:56:15 +00:00
rev = oid_to_hex_r(rev_oid, &new_branch_info->commit->object.oid);
if (opts->checkout_index && opts->checkout_worktree)
patch_mode = ADD_P_CHECKOUT;
else if (opts->checkout_index && !opts->checkout_worktree)
patch_mode = ADD_P_RESET;
else if (!opts->checkout_index && opts->checkout_worktree)
patch_mode = ADD_P_WORKTREE;
else
BUG("either flag must have been set, worktree=%d, index=%d",
opts->checkout_worktree, opts->checkout_index);
return !!run_add_p(the_repository, patch_mode, rev,
&opts->pathspec);
}
repo_hold_locked_index(the_repository, &lock_file, LOCK_DIE_ON_ERROR);
if (repo_read_index_preload(the_repository, &opts->pathspec, 0) < 0)
return error(_("index file corrupt"));
if (opts->source_tree)
read_tree_some(opts->source_tree, &opts->pathspec);
if (opts->merge)
unmerge_index(&the_index, &opts->pathspec, CE_MATCHED);
ps_matched = xcalloc(opts->pathspec.nr, 1);
checkout: avoid unnecessary match_pathspec calls In checkout_paths() we do this - for all updated items, call match_pathspec - for all items, call match_pathspec (inside unmerge_cache) - for all items, call match_pathspec (for showing "path .. is unmerged) - for updated items, call match_pathspec and update paths That's a lot of duplicate match_pathspec(s) and the function is not exactly cheap to be called so many times, especially on large indexes. This patch makes it call match_pathspec once per updated index entry, save the result in ce_flags and reuse the results in the following loops. The changes in 0a1283b (checkout $tree $path: do not clobber local changes in $path not in $tree - 2011-09-30) limit the affected paths to ones we read from $tree. We do not do anything to other modified entries in this case, so the "for all items" above could be modified to "for all updated items". But.. The command's behavior now is modified slightly: unmerged entries that match $path, but not updated by $tree, are now NOT touched. Although this should be considered a bug fix, not a regression. A new test is added for this change. And while at there, free ps_matched after use. The following command is tested on webkit, 215k entries. The pattern is chosen mainly to make match_pathspec sweat: git checkout -- "*[a-zA-Z]*[a-zA-Z]*[a-zA-Z]*" before after real 0m3.493s 0m2.737s user 0m2.239s 0m1.586s sys 0m1.252s 0m1.151s Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-03-27 05:58:21 +00:00
/*
* Make sure all pathspecs participated in locating the paths
* to be checked out.
*/
for (pos = 0; pos < the_index.cache_nr; pos++)
if (opts->overlay_mode)
mark_ce_for_checkout_overlay(the_index.cache[pos],
ps_matched,
opts);
else
mark_ce_for_checkout_no_overlay(the_index.cache[pos],
ps_matched,
opts);
if (report_path_error(ps_matched, &opts->pathspec)) {
checkout: avoid unnecessary match_pathspec calls In checkout_paths() we do this - for all updated items, call match_pathspec - for all items, call match_pathspec (inside unmerge_cache) - for all items, call match_pathspec (for showing "path .. is unmerged) - for updated items, call match_pathspec and update paths That's a lot of duplicate match_pathspec(s) and the function is not exactly cheap to be called so many times, especially on large indexes. This patch makes it call match_pathspec once per updated index entry, save the result in ce_flags and reuse the results in the following loops. The changes in 0a1283b (checkout $tree $path: do not clobber local changes in $path not in $tree - 2011-09-30) limit the affected paths to ones we read from $tree. We do not do anything to other modified entries in this case, so the "for all items" above could be modified to "for all updated items". But.. The command's behavior now is modified slightly: unmerged entries that match $path, but not updated by $tree, are now NOT touched. Although this should be considered a bug fix, not a regression. A new test is added for this change. And while at there, free ps_matched after use. The following command is tested on webkit, 215k entries. The pattern is chosen mainly to make match_pathspec sweat: git checkout -- "*[a-zA-Z]*[a-zA-Z]*[a-zA-Z]*" before after real 0m3.493s 0m2.737s user 0m2.239s 0m1.586s sys 0m1.252s 0m1.151s Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-03-27 05:58:21 +00:00
free(ps_matched);
return 1;
checkout: avoid unnecessary match_pathspec calls In checkout_paths() we do this - for all updated items, call match_pathspec - for all items, call match_pathspec (inside unmerge_cache) - for all items, call match_pathspec (for showing "path .. is unmerged) - for updated items, call match_pathspec and update paths That's a lot of duplicate match_pathspec(s) and the function is not exactly cheap to be called so many times, especially on large indexes. This patch makes it call match_pathspec once per updated index entry, save the result in ce_flags and reuse the results in the following loops. The changes in 0a1283b (checkout $tree $path: do not clobber local changes in $path not in $tree - 2011-09-30) limit the affected paths to ones we read from $tree. We do not do anything to other modified entries in this case, so the "for all items" above could be modified to "for all updated items". But.. The command's behavior now is modified slightly: unmerged entries that match $path, but not updated by $tree, are now NOT touched. Although this should be considered a bug fix, not a regression. A new test is added for this change. And while at there, free ps_matched after use. The following command is tested on webkit, 215k entries. The pattern is chosen mainly to make match_pathspec sweat: git checkout -- "*[a-zA-Z]*[a-zA-Z]*[a-zA-Z]*" before after real 0m3.493s 0m2.737s user 0m2.239s 0m1.586s sys 0m1.252s 0m1.151s Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-03-27 05:58:21 +00:00
}
free(ps_matched);
/* Any unmerged paths? */
for (pos = 0; pos < the_index.cache_nr; pos++) {
const struct cache_entry *ce = the_index.cache[pos];
checkout: avoid unnecessary match_pathspec calls In checkout_paths() we do this - for all updated items, call match_pathspec - for all items, call match_pathspec (inside unmerge_cache) - for all items, call match_pathspec (for showing "path .. is unmerged) - for updated items, call match_pathspec and update paths That's a lot of duplicate match_pathspec(s) and the function is not exactly cheap to be called so many times, especially on large indexes. This patch makes it call match_pathspec once per updated index entry, save the result in ce_flags and reuse the results in the following loops. The changes in 0a1283b (checkout $tree $path: do not clobber local changes in $path not in $tree - 2011-09-30) limit the affected paths to ones we read from $tree. We do not do anything to other modified entries in this case, so the "for all items" above could be modified to "for all updated items". But.. The command's behavior now is modified slightly: unmerged entries that match $path, but not updated by $tree, are now NOT touched. Although this should be considered a bug fix, not a regression. A new test is added for this change. And while at there, free ps_matched after use. The following command is tested on webkit, 215k entries. The pattern is chosen mainly to make match_pathspec sweat: git checkout -- "*[a-zA-Z]*[a-zA-Z]*[a-zA-Z]*" before after real 0m3.493s 0m2.737s user 0m2.239s 0m1.586s sys 0m1.252s 0m1.151s Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-03-27 05:58:21 +00:00
if (ce->ce_flags & CE_MATCHED) {
if (!ce_stage(ce))
continue;
if (opts->ignore_unmerged) {
if (!opts->quiet)
warning(_("path '%s' is unmerged"), ce->name);
} else if (opts->writeout_stage) {
errs |= check_stage(opts->writeout_stage, ce, pos, opts->overlay_mode);
} else if (opts->merge) {
errs |= check_stages((1<<2) | (1<<3), ce, pos);
} else {
errs = 1;
error(_("path '%s' is unmerged"), ce->name);
}
pos = skip_same_name(ce, pos) - 1;
}
}
if (errs)
return 1;
/* Now we are committed to check them out */
if (opts->checkout_worktree)
errs |= checkout_worktree(opts, new_branch_info);
restore: invalidate cache-tree when removing entries with --staged When "git restore --staged <path>" removes a path that's in the index, it marks the entry with CE_REMOVE, but we don't do anything to invalidate the cache-tree. In the non-staged case, we end up in checkout_worktree(), which calls remove_marked_cache_entries(). That actually drops the entries from the index, as well as invalidating the cache-tree and untracked-cache. But with --staged, we never call checkout_worktree(), and the CE_REMOVE entries remain. Interestingly, they are dropped when we write out the index, but that means the resulting index is inconsistent: its cache-tree will not match the actual entries, and running "git commit" immediately after will create the wrong tree. We can solve this by calling remove_marked_cache_entries() ourselves before writing out the index. Note that we can't just hoist it out of checkout_worktree(); that function needs to iterate over the CE_REMOVE entries (to drop their matching worktree files) before removing them. One curiosity about the test: without this patch, it actually triggers a BUG() when running git-restore: BUG: cache-tree.c:810: new1 with flags 0x4420000 should not be in cache-tree But in the original problem report, which used a similar recipe, git-restore actually creates the bogus index (and the commit is created with the wrong tree). I'm not sure why the test here behaves differently than my out-of-suite reproduction, but what's here should catch either symptom (and the fix corrects both cases). Reported-by: Torsten Krah <krah.tm@gmail.com> Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2020-01-08 11:43:44 +00:00
else
remove_marked_cache_entries(&the_index, 1);
/*
* Allow updating the index when checking out from the index.
* This is to save new stat info.
*/
if (opts->checkout_worktree && !opts->checkout_index && !opts->source_tree)
checkout_index = 1;
else
checkout_index = opts->checkout_index;
if (checkout_index) {
if (write_locked_index(&the_index, &lock_file, COMMIT_LOCK))
die(_("unable to write new index file"));
} else {
/*
* NEEDSWORK: if --worktree is not specified, we
* should save stat info of checked out files in the
* index to avoid the next (potentially costly)
* refresh. But it's a bit tricker to do...
*/
rollback_lock_file(&lock_file);
}
read_ref_full("HEAD", 0, &rev, NULL);
head = lookup_commit_reference_gently(the_repository, &rev, 1);
errs |= post_checkout_hook(head, head, 0);
return errs;
}
static void show_local_changes(struct object *head,
const struct diff_options *opts)
{
struct rev_info rev;
/* I think we want full paths, even if we're in a subdirectory. */
repo_init_revisions(the_repository, &rev, NULL);
rev.diffopt.flags = opts->flags;
rev.diffopt.output_format |= DIFF_FORMAT_NAME_STATUS;
rev.diffopt.flags.recursive = 1;
diff_setup_done(&rev.diffopt);
add_pending_object(&rev, head, NULL);
run_diff_index(&rev, 0);
revision.[ch]: provide and start using a release_revisions() The users of the revision.[ch] API's "struct rev_info" are a major source of memory leaks in the test suite under SANITIZE=leak, which in turn adds a lot of noise when trying to mark up tests with "TEST_PASSES_SANITIZE_LEAK=true". The users of that API are largely one-shot, e.g. "git rev-list" or "git log", or the "git checkout" and "git stash" being modified here For these callers freeing the memory is arguably a waste of time, but in many cases they've actually been trying to free the memory, and just doing that in a buggy manner. Let's provide a release_revisions() function for these users, and start migrating them over per the plan outlined in [1]. Right now this only handles the "pending" member of the struct, but more will be added in subsequent commits. Even though we only clear the "pending" member now, let's not leave a trap in code like the pre-image of index_differs_from(), where we'd start doing the wrong thing as soon as the release_revisions() learned to clear its "diffopt". I.e. we need to call release_revisions() after we've inspected any state in "struct rev_info". This leaves in place e.g. clear_pathspec(&rev.prune_data) in stash_working_tree() in builtin/stash.c, subsequent commits will teach release_revisions() to free "prune_data" and other members that in some cases are individually cleared by users of "struct rev_info" by reaching into its members. Those subsequent commits will remove the relevant calls to e.g. clear_pathspec(). We avoid amending code in index_differs_from() in diff-lib.c as well as wt_status_collect_changes_index(), has_unstaged_changes() and has_uncommitted_changes() in wt-status.c in a way that assumes that we are already clearing the "diffopt" member. That will be handled in a subsequent commit. 1. https://lore.kernel.org/git/87a6k8daeu.fsf@evledraar.gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2022-04-13 20:01:35 +00:00
release_revisions(&rev);
}
static void describe_detached_head(const char *msg, struct commit *commit)
{
struct strbuf sb = STRBUF_INIT;
if (!repo_parse_commit(the_repository, commit))
pp_commit_easy(CMIT_FMT_ONELINE, commit, &sb);
if (print_sha1_ellipsis()) {
fprintf(stderr, "%s %s... %s\n", msg,
repo_find_unique_abbrev(the_repository, &commit->object.oid, DEFAULT_ABBREV),
sb.buf);
} else {
fprintf(stderr, "%s %s %s\n", msg,
repo_find_unique_abbrev(the_repository, &commit->object.oid, DEFAULT_ABBREV),
sb.buf);
}
strbuf_release(&sb);
}
static int reset_tree(struct tree *tree, const struct checkout_opts *o,
int worktree, int *writeout_error,
struct branch_info *info)
{
struct unpack_trees_options opts;
struct tree_desc tree_desc;
memset(&opts, 0, sizeof(opts));
opts.head_idx = -1;
opts.update = worktree;
opts.skip_unmerged = !worktree;
2021-09-27 16:33:44 +00:00
opts.reset = o->force ? UNPACK_RESET_OVERWRITE_UNTRACKED :
UNPACK_RESET_PROTECT_UNTRACKED;
opts.preserve_ignored = (!o->force && !o->overwrite_ignore);
opts.merge = 1;
opts.fn = oneway_merge;
opts.verbose_update = o->show_progress;
opts.src_index = &the_index;
opts.dst_index = &the_index;
init_checkout_metadata(&opts.meta, info->refname,
info->commit ? &info->commit->object.oid : null_oid(),
NULL);
if (parse_tree(tree) < 0)
return 128;
init_tree_desc(&tree_desc, &tree->object.oid, tree->buffer, tree->size);
switch (unpack_trees(1, &tree_desc, &opts)) {
case -2:
*writeout_error = 1;
/*
* We return 0 nevertheless, as the index is all right
* and more importantly we have made best efforts to
* update paths in the work tree, and we cannot revert
* them.
*/
consistently use "fallthrough" comments in switches Gcc 7 adds -Wimplicit-fallthrough, which can warn when a switch case falls through to the next case. The general idea is that the compiler can't tell if this was intentional or not, so you should annotate any intentional fall-throughs as such, leaving it to complain about any unannotated ones. There's a GNU __attribute__ which can be used for annotation, but of course we'd have to #ifdef it away on non-gcc compilers. Gcc will also recognize specially-formatted comments, which matches our current practice. Let's extend that practice to all of the unannotated sites (which I did look over and verify that they were behaving as intended). Ideally in each case we'd actually give some reasons in the comment about why we're falling through, or what we're falling through to. And gcc does support that with -Wimplicit-fallthrough=2, which relaxes the comment pattern matching to anything that contains "fallthrough" (or a variety of spelling variants). However, this isn't the default for -Wimplicit-fallthrough, nor for -Wextra. In the name of simplicity, it's probably better for us to support the default level, which requires "fallthrough" to be the only thing in the comment (modulo some window dressing like "else" and some punctuation; see the gcc manual for the complete set of patterns). This patch suppresses all warnings due to -Wimplicit-fallthrough. We might eventually want to add that to the DEVELOPER Makefile knob, but we should probably wait until gcc 7 is more widely adopted (since earlier versions will complain about the unknown warning type). Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-09-21 06:25:41 +00:00
/* fallthrough */
case 0:
return 0;
default:
return 128;
}
}
static void setup_branch_path(struct branch_info *branch)
{
struct strbuf buf = STRBUF_INIT;
/*
* If this is a ref, resolve it; otherwise, look up the OID for our
* expression. Failure here is okay.
*/
if (!repo_dwim_ref(the_repository, branch->name, strlen(branch->name),
&branch->oid, &branch->refname, 0))
repo_get_oid_committish(the_repository, branch->name, &branch->oid);
strbuf_branchname(&buf, branch->name, INTERPRET_BRANCH_LOCAL);
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
if (strcmp(buf.buf, branch->name)) {
free(branch->name);
branch->name = xstrdup(buf.buf);
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
}
strbuf_splice(&buf, 0, 0, "refs/heads/", 11);
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
free(branch->path);
branch->path = strbuf_detach(&buf, NULL);
}
static void init_topts(struct unpack_trees_options *topts, int merge,
int show_progress, int overwrite_ignore,
struct commit *old_commit)
{
memset(topts, 0, sizeof(*topts));
topts->head_idx = -1;
topts->src_index = &the_index;
topts->dst_index = &the_index;
setup_unpack_trees_porcelain(topts, "checkout");
topts->initial_checkout = is_index_unborn(&the_index);
topts->update = 1;
topts->merge = 1;
topts->quiet = merge && old_commit;
topts->verbose_update = show_progress;
topts->fn = twoway_merge;
topts->preserve_ignored = !overwrite_ignore;
}
static int merge_working_tree(const struct checkout_opts *opts,
struct branch_info *old_branch_info,
struct branch_info *new_branch_info,
int *writeout_error)
{
int ret;
struct lock_file lock_file = LOCK_INIT;
struct tree *new_tree;
repo_hold_locked_index(the_repository, &lock_file, LOCK_DIE_ON_ERROR);
if (repo_read_index_preload(the_repository, NULL, 0) < 0)
return error(_("index file corrupt"));
resolve_undo_clear_index(&the_index);
if (opts->new_orphan_branch && opts->orphan_from_empty_tree) {
if (new_branch_info->commit)
BUG("'switch --orphan' should never accept a commit as starting point");
new_tree = parse_tree_indirect(the_hash_algo->empty_tree);
if (!new_tree)
BUG("unable to read empty tree");
} else {
new_tree = repo_get_commit_tree(the_repository,
new_branch_info->commit);
if (!new_tree)
return error(_("unable to read tree (%s)"),
oid_to_hex(&new_branch_info->commit->object.oid));
}
if (opts->discard_changes) {
ret = reset_tree(new_tree, opts, 1, writeout_error, new_branch_info);
if (ret)
return ret;
} else {
struct tree_desc trees[2];
struct tree *tree;
struct unpack_trees_options topts;
const struct object_id *old_commit_oid;
refresh_index(&the_index, REFRESH_QUIET, NULL, NULL, NULL);
if (unmerged_index(&the_index)) {
error(_("you need to resolve your current index first"));
return 1;
}
/* 2-way merge to the new branch */
init_topts(&topts, opts->merge, opts->show_progress,
opts->overwrite_ignore, old_branch_info->commit);
init_checkout_metadata(&topts.meta, new_branch_info->refname,
new_branch_info->commit ?
&new_branch_info->commit->object.oid :
&new_branch_info->oid, NULL);
old_commit_oid = old_branch_info->commit ?
&old_branch_info->commit->object.oid :
the_hash_algo->empty_tree;
tree = parse_tree_indirect(old_commit_oid);
if (!tree)
die(_("unable to parse commit %s"),
oid_to_hex(old_commit_oid));
init_tree_desc(&trees[0], &tree->object.oid,
tree->buffer, tree->size);
if (parse_tree(new_tree) < 0)
exit(128);
tree = new_tree;
init_tree_desc(&trees[1], &tree->object.oid,
tree->buffer, tree->size);
ret = unpack_trees(2, trees, &topts);
clear_unpack_trees_porcelain(&topts);
if (ret == -1) {
/*
* Unpack couldn't do a trivial merge; either
* give up or do a real merge, depending on
* whether the merge flag was used.
*/
struct tree *work;
checkout: prevent losing staged changes with --merge When --merge is specified, we may need to do a real merge (instead of three-way tree unpacking), the steps are best seen in git-checkout.sh version before it's removed: # Match the index to the working tree, and do a three-way. git diff-files --name-only | git update-index --remove --stdin && work=`git write-tree` && git read-tree $v --reset -u $new || exit git merge-recursive $old -- $new $work # Do not register the cleanly merged paths in the index yet. # this is not a real merge before committing, but just carrying # the working tree changes along. unmerged=`git ls-files -u` git read-tree $v --reset $new case "$unmerged" in '') ;; *) ( z40=0000000000000000000000000000000000000000 echo "$unmerged" | sed -e 's/^[0-7]* [0-9a-f]* /'"0 $z40 /" echo "$unmerged" ) | git update-index --index-info ;; esac Notice the last 'read-tree --reset' step. We restore worktree back to 'new' tree after worktree's messed up by merge-recursive. If there are staged changes before this whole command sequence is executed, they are lost because they are unlikely part of the 'new' tree to be restored. There is no easy way to fix this. Elijah may have something up his sleeves [1], but until then, check if there are staged changes and refuse to run and lose them. The user would need to do "git reset" to continue in this case. A note about the test update. 'checkout -m' in that test will fail because a deletion is staged. This 'checkout -m' was previously needed to verify quietness behavior of unpack-trees. But a different check has been put in place in the last patch. We can safely drop 'checkout -m' now. [1] CABPp-BFoL_U=bzON4SEMaQSKU2TKwnOgNqjt5MUaOejTKGUJxw@mail.gmail.com Reported-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-03-22 09:31:38 +00:00
struct tree *old_tree;
struct merge_options o;
2019-03-19 09:39:10 +00:00
struct strbuf sb = STRBUF_INIT;
struct strbuf old_commit_shortname = STRBUF_INIT;
2019-03-19 09:39:10 +00:00
if (!opts->merge)
return 1;
/*
* Without old_branch_info->commit, the below is the same as
* the two-tree unpack we already tried and failed.
*/
if (!old_branch_info->commit)
return 1;
old_tree = repo_get_commit_tree(the_repository,
old_branch_info->commit);
checkout: prevent losing staged changes with --merge When --merge is specified, we may need to do a real merge (instead of three-way tree unpacking), the steps are best seen in git-checkout.sh version before it's removed: # Match the index to the working tree, and do a three-way. git diff-files --name-only | git update-index --remove --stdin && work=`git write-tree` && git read-tree $v --reset -u $new || exit git merge-recursive $old -- $new $work # Do not register the cleanly merged paths in the index yet. # this is not a real merge before committing, but just carrying # the working tree changes along. unmerged=`git ls-files -u` git read-tree $v --reset $new case "$unmerged" in '') ;; *) ( z40=0000000000000000000000000000000000000000 echo "$unmerged" | sed -e 's/^[0-7]* [0-9a-f]* /'"0 $z40 /" echo "$unmerged" ) | git update-index --index-info ;; esac Notice the last 'read-tree --reset' step. We restore worktree back to 'new' tree after worktree's messed up by merge-recursive. If there are staged changes before this whole command sequence is executed, they are lost because they are unlikely part of the 'new' tree to be restored. There is no easy way to fix this. Elijah may have something up his sleeves [1], but until then, check if there are staged changes and refuse to run and lose them. The user would need to do "git reset" to continue in this case. A note about the test update. 'checkout -m' in that test will fail because a deletion is staged. This 'checkout -m' was previously needed to verify quietness behavior of unpack-trees. But a different check has been put in place in the last patch. We can safely drop 'checkout -m' now. [1] CABPp-BFoL_U=bzON4SEMaQSKU2TKwnOgNqjt5MUaOejTKGUJxw@mail.gmail.com Reported-by: Phillip Wood <phillip.wood@dunelm.org.uk> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-03-22 09:31:38 +00:00
if (repo_index_has_changes(the_repository, old_tree, &sb))
die(_("cannot continue with staged changes in "
"the following files:\n%s"), sb.buf);
strbuf_release(&sb);
/* Do more real merge */
/*
* We update the index fully, then write the
* tree from the index, then merge the new
* branch with the current tree, with the old
* branch as the base. Then we reset the index
* (but not the working tree) to the new
* branch, leaving the working tree as the
* merged version, but skipping unmerged
* entries in the index.
*/
add_files_to_cache(the_repository, NULL, NULL, NULL, 0,
0);
init_merge_options(&o, the_repository);
o.verbosity = 0;
work = write_in_core_index_as_tree(the_repository);
ret = reset_tree(new_tree,
opts, 1,
writeout_error, new_branch_info);
if (ret)
return ret;
o.ancestor = old_branch_info->name;
if (!old_branch_info->name) {
strbuf_add_unique_abbrev(&old_commit_shortname,
&old_branch_info->commit->object.oid,
DEFAULT_ABBREV);
o.ancestor = old_commit_shortname.buf;
}
o.branch1 = new_branch_info->name;
o.branch2 = "local";
o.conflict_style = opts->conflict_style;
ret = merge_trees(&o,
new_tree,
work,
old_tree);
prepare the builtins for a libified merge_recursive() Previously, callers of merge_trees() or merge_recursive() expected that code to die() with an error message. This used to be okay because we called those commands from scripts, and had a chance to print out a message in case the command failed fatally (read: with exit code 128). As scripting incurs its own set of problems (portability, speed, idiosyncrasies of different shells, limited data structures leading to inefficient code), we are converting more and more of these scripts into builtins, using library functions directly. We already tried to use merge_recursive() directly in the builtin git-am, for example. Unfortunately, we had to roll it back temporarily because some of the code in merge-recursive.c still deemed it okay to call die(), when the builtin am code really wanted to print out a useful advice after the merge failed fatally. In the next commits, we want to fix that. The code touched by this commit expected merge_trees() to die() with some useful message when there is an error condition, but merge_trees() is going to be improved by converting all die() calls to return error() instead (i.e. return value -1 after printing out the message as before), so that the caller can react more flexibly. This is a step to prepare for the version of merge_trees() that no longer dies, even if we just imitate the previous behavior by calling exit(128): this is what callers of e.g. `git merge` have come to expect. Note that the callers of the sequencer (revert and cherry-pick) already fail fast even for the return value -1; The only difference is that they now get a chance to say "<command> failed". A caller of merge_trees() might want handle error messages themselves (or even suppress them). As this patch is already complex enough, we leave that change for a later patch. Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2016-07-26 16:06:02 +00:00
if (ret < 0)
exit(128);
ret = reset_tree(new_tree,
opts, 0,
writeout_error, new_branch_info);
strbuf_release(&o.obuf);
strbuf_release(&old_commit_shortname);
if (ret)
return ret;
}
}
if (!cache_tree_fully_valid(the_index.cache_tree))
cache_tree_update(&the_index, WRITE_TREE_SILENT | WRITE_TREE_REPAIR);
if (write_locked_index(&the_index, &lock_file, COMMIT_LOCK))
die(_("unable to write new index file"));
if (!opts->discard_changes && !opts->quiet && new_branch_info->commit)
show_local_changes(&new_branch_info->commit->object, &opts->diff_options);
return 0;
}
static void report_tracking(struct branch_info *new_branch_info)
{
struct strbuf sb = STRBUF_INIT;
struct branch *branch = branch_get(new_branch_info->name);
if (!format_tracking_info(branch, &sb, AHEAD_BEHIND_FULL, 1))
return;
fputs(sb.buf, stdout);
strbuf_release(&sb);
}
static void update_refs_for_switch(const struct checkout_opts *opts,
struct branch_info *old_branch_info,
struct branch_info *new_branch_info)
{
struct strbuf msg = STRBUF_INIT;
const char *old_desc, *reflog_msg;
if (opts->new_branch) {
if (opts->new_orphan_branch) {
char *refname;
refname = mkpathdup("refs/heads/%s", opts->new_orphan_branch);
if (opts->new_branch_log &&
!should_autocreate_reflog(refname)) {
int ret;
struct strbuf err = STRBUF_INIT;
ret = safe_create_reflog(refname, &err);
if (ret) {
fprintf(stderr, _("Can not do reflog for '%s': %s\n"),
opts->new_orphan_branch, err.buf);
strbuf_release(&err);
free(refname);
return;
}
strbuf_release(&err);
}
free(refname);
}
else
create_branch(the_repository,
opts->new_branch, new_branch_info->name,
opts->new_branch_force ? 1 : 0,
opts->new_branch_force ? 1 : 0,
opts->new_branch_log,
opts->quiet,
opts->track,
0);
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
free(new_branch_info->name);
free(new_branch_info->refname);
new_branch_info->name = xstrdup(opts->new_branch);
setup_branch_path(new_branch_info);
}
old_desc = old_branch_info->name;
if (!old_desc && old_branch_info->commit)
old_desc = oid_to_hex(&old_branch_info->commit->object.oid);
reflog_msg = getenv("GIT_REFLOG_ACTION");
if (!reflog_msg)
strbuf_addf(&msg, "checkout: moving from %s to %s",
old_desc ? old_desc : "(invalid)", new_branch_info->name);
else
strbuf_insertstr(&msg, 0, reflog_msg);
if (!strcmp(new_branch_info->name, "HEAD") && !new_branch_info->path && !opts->force_detach) {
/* Nothing to do. */
} else if (opts->force_detach || !new_branch_info->path) { /* No longer on any branch. */
update_ref(msg.buf, "HEAD", &new_branch_info->commit->object.oid, NULL,
REF_NO_DEREF, UPDATE_REFS_DIE_ON_ERR);
if (!opts->quiet) {
if (old_branch_info->path &&
advice_enabled(ADVICE_DETACHED_HEAD) && !opts->force_detach)
detach_advice(new_branch_info->name);
describe_detached_head(_("HEAD is now at"), new_branch_info->commit);
}
} else if (new_branch_info->path) { /* Switch branches. */
if (create_symref("HEAD", new_branch_info->path, msg.buf) < 0)
die(_("unable to update HEAD"));
if (!opts->quiet) {
if (old_branch_info->path && !strcmp(new_branch_info->path, old_branch_info->path)) {
if (opts->new_branch_force)
fprintf(stderr, _("Reset branch '%s'\n"),
new_branch_info->name);
else
fprintf(stderr, _("Already on '%s'\n"),
new_branch_info->name);
} else if (opts->new_branch) {
if (opts->branch_exists)
fprintf(stderr, _("Switched to and reset branch '%s'\n"), new_branch_info->name);
else
fprintf(stderr, _("Switched to a new branch '%s'\n"), new_branch_info->name);
} else {
fprintf(stderr, _("Switched to branch '%s'\n"),
new_branch_info->name);
}
}
if (old_branch_info->path && old_branch_info->name) {
if (!ref_exists(old_branch_info->path) && reflog_exists(old_branch_info->path))
delete_reflog(old_branch_info->path);
}
}
remove_branch_state(the_repository, !opts->quiet);
strbuf_release(&msg);
if (!opts->quiet &&
checkout: omit "tracking" information on a detached HEAD By definition, a detached HEAD state is tentative and there is no configured "upstream" that it always wants to integrate with. But if you detach from a branch that is behind its upstream, e.g., $ git checkout -t -b main origin/main $ git checkout main $ git reset --hard HEAD^ $ git checkout --detach main you'd see "you are behind your upstream origin/main". This does not happen when you replace the last step in the above with any of these $ git checkout HEAD^0 $ git checkout --detach HEAD $ git checkout --detach origin/main Before 32669671 (checkout: introduce --detach synonym for "git checkout foo^{commit}", 2011-02-08) introduced the "--detach" option, the rule to decide if we show the tracking information used to be: If --quiet is not given, and if the given branch name is a real local branch (i.e. the one we can compute the file path under .git/, like 'refs/heads/master' or "HEAD" which stand for the name of the current branch", then give the tracking information. to exclude things like "git checkout master^0" (which was the official way to detach HEAD at the commit before that commit) and "git checkout origin/master^0" from showing tracking information, but still do show the tracking information for the current branch for "git checkout HEAD". The introduction of an explicit option "--detach" broke this subtley. The new rule should have been If --quiet is given, do not bother with tracking info. If --detach is given, do not bother with tracking info. Otherwise, if we know that the branch name given is a real local branch, or if we were given "HEAD" and "HEAD" is not detached, then attempt to show the tracking info. but it allowed "git checkout --detach master" to also show the tracking info by mistake. Let's tighten the rule to fix this. Reported-by: mirth hickford <mirth.hickford@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2024-03-30 21:37:14 +00:00
!opts->force_detach &&
(new_branch_info->path || !strcmp(new_branch_info->name, "HEAD")))
report_tracking(new_branch_info);
}
static int add_pending_uninteresting_ref(const char *refname,
const struct object_id *oid,
int flags UNUSED, void *cb_data)
{
add_pending_oid(cb_data, refname, oid, UNINTERESTING);
return 0;
}
static void describe_one_orphan(struct strbuf *sb, struct commit *commit)
{
strbuf_addstr(sb, " ");
strbuf_add_unique_abbrev(sb, &commit->object.oid, DEFAULT_ABBREV);
strbuf_addch(sb, ' ');
if (!repo_parse_commit(the_repository, commit))
pp_commit_easy(CMIT_FMT_ONELINE, commit, sb);
strbuf_addch(sb, '\n');
}
#define ORPHAN_CUTOFF 4
static void suggest_reattach(struct commit *commit, struct rev_info *revs)
{
struct commit *c, *last = NULL;
struct strbuf sb = STRBUF_INIT;
int lost = 0;
while ((c = get_revision(revs)) != NULL) {
if (lost < ORPHAN_CUTOFF)
describe_one_orphan(&sb, c);
last = c;
lost++;
}
if (ORPHAN_CUTOFF < lost) {
int more = lost - ORPHAN_CUTOFF;
if (more == 1)
describe_one_orphan(&sb, last);
else
strbuf_addf(&sb, _(" ... and %d more.\n"), more);
}
fprintf(stderr,
Q_(
/* The singular version */
"Warning: you are leaving %d commit behind, "
"not connected to\n"
"any of your branches:\n\n"
"%s\n",
/* The plural version */
"Warning: you are leaving %d commits behind, "
"not connected to\n"
"any of your branches:\n\n"
"%s\n",
/* Give ngettext() the count */
lost),
lost,
sb.buf);
strbuf_release(&sb);
if (advice_enabled(ADVICE_DETACHED_HEAD))
fprintf(stderr,
Q_(
/* The singular version */
"If you want to keep it by creating a new branch, "
"this may be a good time\nto do so with:\n\n"
" git branch <new-branch-name> %s\n\n",
/* The plural version */
"If you want to keep them by creating a new branch, "
"this may be a good time\nto do so with:\n\n"
" git branch <new-branch-name> %s\n\n",
/* Give ngettext() the count */
lost),
repo_find_unique_abbrev(the_repository, &commit->object.oid, DEFAULT_ABBREV));
}
/*
* We are about to leave commit that was at the tip of a detached
* HEAD. If it is not reachable from any ref, this is the last chance
* for the user to do so without resorting to reflog.
*/
static void orphaned_commit_warning(struct commit *old_commit, struct commit *new_commit)
{
struct rev_info revs;
struct object *object = &old_commit->object;
repo_init_revisions(the_repository, &revs, NULL);
setup_revisions(0, NULL, &revs, NULL);
object->flags &= ~UNINTERESTING;
add_pending_object(&revs, object, oid_to_hex(&object->oid));
for_each_ref(add_pending_uninteresting_ref, &revs);
if (new_commit)
add_pending_oid(&revs, "HEAD",
&new_commit->object.oid,
UNINTERESTING);
if (prepare_revision_walk(&revs))
die(_("internal error in revision walk"));
if (!(old_commit->object.flags & UNINTERESTING))
suggest_reattach(old_commit, &revs);
else
describe_detached_head(_("Previous HEAD position was"), old_commit);
/* Clean up objects used, as they will be reused. */
repo_clear_commit_marks(the_repository, ALL_REV_FLAGS);
release_revisions(&revs);
}
static int switch_branches(const struct checkout_opts *opts,
struct branch_info *new_branch_info)
{
int ret = 0;
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
struct branch_info old_branch_info = { 0 };
struct object_id rev;
int flag, writeout_error = 0;
int do_merge = 1;
trace2_cmd_mode("branch");
memset(&old_branch_info, 0, sizeof(old_branch_info));
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
old_branch_info.path = resolve_refdup("HEAD", 0, &rev, &flag);
if (old_branch_info.path)
old_branch_info.commit = lookup_commit_reference_gently(the_repository, &rev, 1);
if (!(flag & REF_ISSYMREF))
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
FREE_AND_NULL(old_branch_info.path);
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
if (old_branch_info.path) {
const char *const prefix = "refs/heads/";
const char *p;
if (skip_prefix(old_branch_info.path, prefix, &p))
old_branch_info.name = xstrdup(p);
}
if (opts->new_orphan_branch && opts->orphan_from_empty_tree) {
if (new_branch_info->name)
BUG("'switch --orphan' should never accept a commit as starting point");
new_branch_info->commit = NULL;
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
new_branch_info->name = xstrdup("(empty)");
do_merge = 1;
}
if (!new_branch_info->name) {
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
new_branch_info->name = xstrdup("HEAD");
new_branch_info->commit = old_branch_info.commit;
if (!new_branch_info->commit)
die(_("You are on a branch yet to be born"));
parse_commit_or_die(new_branch_info->commit);
if (opts->only_merge_on_switching_branches)
do_merge = 0;
}
if (do_merge) {
ret = merge_working_tree(opts, &old_branch_info, new_branch_info, &writeout_error);
if (ret) {
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
branch_info_release(&old_branch_info);
return ret;
}
}
if (!opts->quiet && !old_branch_info.path && old_branch_info.commit && new_branch_info->commit != old_branch_info.commit)
orphaned_commit_warning(old_branch_info.commit, new_branch_info->commit);
update_refs_for_switch(opts, &old_branch_info, new_branch_info);
ret = post_checkout_hook(old_branch_info.commit, new_branch_info->commit, 1);
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
branch_info_release(&old_branch_info);
return ret || writeout_error;
}
config: add ctx arg to config_fn_t Add a new "const struct config_context *ctx" arg to config_fn_t to hold additional information about the config iteration operation. config_context has a "struct key_value_info kvi" member that holds metadata about the config source being read (e.g. what kind of config source it is, the filename, etc). In this series, we're only interested in .kvi, so we could have just used "struct key_value_info" as an arg, but config_context makes it possible to add/adjust members in the future without changing the config_fn_t signature. We could also consider other ways of organizing the args (e.g. moving the config name and value into config_context or key_value_info), but in my experiments, the incremental benefit doesn't justify the added complexity (e.g. a config_fn_t will sometimes invoke another config_fn_t but with a different config value). In subsequent commits, the .kvi member will replace the global "struct config_reader" in config.c, making config iteration a global-free operation. It requires much more work for the machinery to provide meaningful values of .kvi, so for now, merely change the signature and call sites, pass NULL as a placeholder value, and don't rely on the arg in any meaningful way. Most of the changes are performed by contrib/coccinelle/config_fn_ctx.pending.cocci, which, for every config_fn_t: - Modifies the signature to accept "const struct config_context *ctx" - Passes "ctx" to any inner config_fn_t, if needed - Adds UNUSED attributes to "ctx", if needed Most config_fn_t instances are easily identified by seeing if they are called by the various config functions. Most of the remaining ones are manually named in the .cocci patch. Manual cleanups are still needed, but the majority of it is trivial; it's either adjusting config_fn_t that the .cocci patch didn't catch, or adding forward declarations of "struct config_context ctx" to make the signatures make sense. The non-trivial changes are in cases where we are invoking a config_fn_t outside of config machinery, and we now need to decide what value of "ctx" to pass. These cases are: - trace2/tr2_cfg.c:tr2_cfg_set_fl() This is indirectly called by git_config_set() so that the trace2 machinery can notice the new config values and update its settings using the tr2 config parsing function, i.e. tr2_cfg_cb(). - builtin/checkout.c:checkout_main() This calls git_xmerge_config() as a shorthand for parsing a CLI arg. This might be worth refactoring away in the future, since git_xmerge_config() can call git_default_config(), which can do much more than just parsing. Handle them by creating a KVI_INIT macro that initializes "struct key_value_info" to a reasonable default, and use that to construct the "ctx" arg. Signed-off-by: Glen Choo <chooglen@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-06-28 19:26:22 +00:00
static int git_checkout_config(const char *var, const char *value,
const struct config_context *ctx, void *cb)
{
struct checkout_opts *opts = cb;
if (!strcmp(var, "diff.ignoresubmodules")) {
if (!value)
return config_error_nonbool(var);
handle_ignore_submodules_arg(&opts->diff_options, value);
return 0;
}
if (!strcmp(var, "checkout.guess")) {
opts->dwim_new_local_branch = git_config_bool(var, value);
return 0;
}
if (starts_with(var, "submodule."))
return git_default_submodule_config(var, value, NULL);
config: add ctx arg to config_fn_t Add a new "const struct config_context *ctx" arg to config_fn_t to hold additional information about the config iteration operation. config_context has a "struct key_value_info kvi" member that holds metadata about the config source being read (e.g. what kind of config source it is, the filename, etc). In this series, we're only interested in .kvi, so we could have just used "struct key_value_info" as an arg, but config_context makes it possible to add/adjust members in the future without changing the config_fn_t signature. We could also consider other ways of organizing the args (e.g. moving the config name and value into config_context or key_value_info), but in my experiments, the incremental benefit doesn't justify the added complexity (e.g. a config_fn_t will sometimes invoke another config_fn_t but with a different config value). In subsequent commits, the .kvi member will replace the global "struct config_reader" in config.c, making config iteration a global-free operation. It requires much more work for the machinery to provide meaningful values of .kvi, so for now, merely change the signature and call sites, pass NULL as a placeholder value, and don't rely on the arg in any meaningful way. Most of the changes are performed by contrib/coccinelle/config_fn_ctx.pending.cocci, which, for every config_fn_t: - Modifies the signature to accept "const struct config_context *ctx" - Passes "ctx" to any inner config_fn_t, if needed - Adds UNUSED attributes to "ctx", if needed Most config_fn_t instances are easily identified by seeing if they are called by the various config functions. Most of the remaining ones are manually named in the .cocci patch. Manual cleanups are still needed, but the majority of it is trivial; it's either adjusting config_fn_t that the .cocci patch didn't catch, or adding forward declarations of "struct config_context ctx" to make the signatures make sense. The non-trivial changes are in cases where we are invoking a config_fn_t outside of config machinery, and we now need to decide what value of "ctx" to pass. These cases are: - trace2/tr2_cfg.c:tr2_cfg_set_fl() This is indirectly called by git_config_set() so that the trace2 machinery can notice the new config values and update its settings using the tr2 config parsing function, i.e. tr2_cfg_cb(). - builtin/checkout.c:checkout_main() This calls git_xmerge_config() as a shorthand for parsing a CLI arg. This might be worth refactoring away in the future, since git_xmerge_config() can call git_default_config(), which can do much more than just parsing. Handle them by creating a KVI_INIT macro that initializes "struct key_value_info" to a reasonable default, and use that to construct the "ctx" arg. Signed-off-by: Glen Choo <chooglen@google.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2023-06-28 19:26:22 +00:00
return git_xmerge_config(var, value, ctx, NULL);
}
static void setup_new_branch_info_and_source_tree(
struct branch_info *new_branch_info,
struct checkout_opts *opts,
struct object_id *rev,
const char *arg)
{
struct tree **source_tree = &opts->source_tree;
struct object_id branch_rev;
add-patch: classify '@' as a synonym for 'HEAD' Currently, (restore, checkout, reset) commands correctly take '@' as a synonym for 'HEAD'. However, in patch mode different prompts/messages are given on command line due to patch mode machinery not considering '@' to be a synonym for 'HEAD' due to literal string comparison with the word 'HEAD', and therefore assigning patch_mode_($command)_nothead and triggering reverse mode (-R in diff-index). The NEEDSWORK comment suggested comparing commit objects to get around this. However, doing so would also take a non-checked out branch pointing to the same commit as HEAD, as HEAD. This would cause confusion to the user. Therefore, after parsing '@', replace it with 'HEAD' as reasonably early as possible. This also solves another problem of disparity between 'git checkout HEAD' and 'git checkout @' (latter detaches at the HEAD commit and the former does not). Trade-offs: - Some of the errors would show the revision argument as 'HEAD' when given '@'. This should be fine, as most users who probably use '@' would be aware that it is a shortcut for 'HEAD' and most probably used to use 'HEAD'. There is also relevant documentation in 'gitrevisions' manpage about '@' being the shortcut for 'HEAD'. Also, the simplicity of the solution far outweighs this cost. - Consider '@' as a shortcut for 'HEAD' even if 'refs/heads/@' exists at a different commit. Naming a branch '@' is an obvious foot-gun and many existing commands already take '@' for 'HEAD' even if 'refs/heads/@' exists at a different commit or does not exist at all (e.g. 'git log @', 'git push origin @' etc.). Therefore this is an existing assumption and should not be a problem. Helped-by: Junio C Hamano <gitster@pobox.com> Helped-by: Phillip Wood <phillip.wood123@gmail.com> Signed-off-by: Ghanshyam Thakkar <shyamthakkar001@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2024-02-13 00:05:29 +00:00
/* treat '@' as a shortcut for 'HEAD' */
new_branch_info->name = !strcmp(arg, "@") ? xstrdup("HEAD") :
xstrdup(arg);
setup_branch_path(new_branch_info);
if (!check_refname_format(new_branch_info->path, 0) &&
!read_ref(new_branch_info->path, &branch_rev))
oidcpy(rev, &branch_rev);
checkout: fix "branch info" memory leaks The "checkout" command is one of the main sources of leaks in the test suite, let's fix the common ones by not leaking from the "struct branch_info". Doing this is rather straightforward, albeit verbose, we need to xstrdup() constant strings going into the struct, and free() the ones we clobber as we go along. This also means that we can delete previous partial leak fixes in this area, i.e. the "path_to_free" accounting added by 96ec7b1e708 (Convert resolve_ref+xstrdup to new resolve_refdup function, 2011-12-13). There was some discussion about whether "we should retain the "const char *" here and cast at free() time, or have it be a "char *". Since this is not a public API with any sort of API boundary let's use "char *", as is already being done for the "refname" member of the same struct. The tests to mark as passing were found with: rm .prove; GIT_SKIP_TESTS=t0027 prove -j8 --state=save t[0-9]*.sh :: --immediate # apply & compile this change prove -j8 --state=failed :: --immediate I.e. the ones that were newly passing when the --state=failed command was run. I left out "t3040-subprojects-basic.sh" and "t4131-apply-fake-ancestor.sh" to to optimization-level related differences similar to the ones noted in[1], except that these would be something the current 'linux-leaks' job would run into. 1. https://lore.kernel.org/git/cover-v3-0.6-00000000000-20211022T175227Z-avarab@gmail.com/ Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-11-16 18:27:38 +00:00
else
/* not an existing branch */
FREE_AND_NULL(new_branch_info->path);
new_branch_info->commit = lookup_commit_reference_gently(the_repository, rev, 1);
if (!new_branch_info->commit) {
/* not a commit */
*source_tree = parse_tree_indirect(rev);
if (!*source_tree)
die(_("unable to read tree (%s)"), oid_to_hex(rev));
} else {
parse_commit_or_die(new_branch_info->commit);
*source_tree = repo_get_commit_tree(the_repository,
new_branch_info->commit);
if (!*source_tree)
die(_("unable to read tree (%s)"),
oid_to_hex(&new_branch_info->commit->object.oid));
}
}
static const char *parse_remote_branch(const char *arg,
struct object_id *rev,
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
int could_be_checkout_paths)
{
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
int num_matches = 0;
const char *remote = unique_tracking_name(arg, rev, &num_matches);
if (remote && could_be_checkout_paths) {
die(_("'%s' could be both a local file and a tracking branch.\n"
"Please use -- (and optionally --no-guess) to disambiguate"),
arg);
}
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
if (!remote && num_matches > 1) {
if (advice_enabled(ADVICE_CHECKOUT_AMBIGUOUS_REMOTE_BRANCH_NAME)) {
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
advise(_("If you meant to check out a remote tracking branch on, e.g. 'origin',\n"
"you can do so by fully qualifying the name with the --track option:\n"
"\n"
" git checkout --track origin/<name>\n"
"\n"
"If you'd like to always have checkouts of an ambiguous <name> prefer\n"
"one remote, e.g. the 'origin' remote, consider setting\n"
"checkout.defaultRemote=origin in your config."));
}
die(_("'%s' matched multiple (%d) remote tracking branches"),
arg, num_matches);
}
return remote;
}
static int parse_branchname_arg(int argc, const char **argv,
int dwim_new_local_branch_ok,
struct branch_info *new_branch_info,
struct checkout_opts *opts,
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
struct object_id *rev)
{
const char **new_branch = &opts->new_branch;
int argcount = 0;
const char *arg;
int dash_dash_pos;
int has_dash_dash = 0;
int i;
/*
* case 1: git checkout <ref> -- [<paths>]
*
* <ref> must be a valid tree, everything after the '--' must be
* a path.
*
* case 2: git checkout -- [<paths>]
*
* everything after the '--' must be paths.
*
* case 3: git checkout <something> [--]
*
* (a) If <something> is a commit, that is to
* switch to the branch or detach HEAD at it. As a special case,
* if <something> is A...B (missing A or B means HEAD but you can
* omit at most one side), and if there is a unique merge base
* between A and B, A...B names that merge base.
*
* (b) If <something> is _not_ a commit, either "--" is present
* or <something> is not a path, no -t or -b was given,
* and there is a tracking branch whose name is <something>
checkout & worktree: introduce checkout.defaultRemote Introduce a checkout.defaultRemote setting which can be used to designate a remote to prefer (via checkout.defaultRemote=origin) when running e.g. "git checkout master" to mean origin/master, even though there's other remotes that have the "master" branch. I want this because it's very handy to use this workflow to checkout a repository and create a topic branch, then get back to a "master" as retrieved from upstream: ( cd /tmp && rm -rf tbdiff && git clone git@github.com:trast/tbdiff.git && cd tbdiff && git branch -m topic && git checkout master ) That will output: Branch 'master' set up to track remote branch 'master' from 'origin'. Switched to a new branch 'master' But as soon as a new remote is added (e.g. just to inspect something from someone else) the DWIMery goes away: ( cd /tmp && rm -rf tbdiff && git clone git@github.com:trast/tbdiff.git && cd tbdiff && git branch -m topic && git remote add avar git@github.com:avar/tbdiff.git && git fetch avar && git checkout master ) Will output (without the advice output added earlier in this series): error: pathspec 'master' did not match any file(s) known to git. The new checkout.defaultRemote config allows me to say that whenever that ambiguity comes up I'd like to prefer "origin", and it'll still work as though the only remote I had was "origin". Also adjust the advice.checkoutAmbiguousRemoteBranchName message to mention this new config setting to the user, the full output on my git.git is now (the last paragraph is new): $ ./git --exec-path=$PWD checkout master error: pathspec 'master' did not match any file(s) known to git. hint: 'master' matched more than one remote tracking branch. hint: We found 26 remotes with a reference that matched. So we fell back hint: on trying to resolve the argument as a path, but failed there too! hint: hint: If you meant to check out a remote tracking branch on, e.g. 'origin', hint: you can do so by fully qualifying the name with the --track option: hint: hint: git checkout --track origin/<name> hint: hint: If you'd like to always have checkouts of an ambiguous <name> prefer hint: one remote, e.g. the 'origin' remote, consider setting hint: checkout.defaultRemote=origin in your config. I considered splitting this into checkout.defaultRemote and worktree.defaultRemote, but it's probably less confusing to break our own rules that anything shared between config should live in core.* than have two config settings, and I couldn't come up with a short name under core.* that made sense (core.defaultRemoteForCheckout?). See also 70c9ac2f19 ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"", 2009-10-18) which introduced this DWIM feature to begin with, and 4e85333197 ("worktree: make add <path> <branch> dwim", 2017-11-26) which added it to git-worktree. Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-06-05 14:40:49 +00:00
* in one and only one remote (or if the branch exists on the
* remote named in checkout.defaultRemote), then this is a
* short-hand to fork local <something> from that
* remote-tracking branch.
*
* (c) Otherwise, if "--" is present, treat it like case (1).
*
* (d) Otherwise :
* - if it's a reference, treat it like case (1)
* - else if it's a path, treat it like case (2)
* - else: fail.
*
* case 4: git checkout <something> <paths>
*
* The first argument must not be ambiguous.
* - If it's *only* a reference, treat it like case (1).
* - If it's only a path, treat it like case (2).
* - else: fail.
*
*/
if (!argc)
return 0;
if (!opts->accept_pathspec) {
if (argc > 1)
die(_("only one reference expected"));
has_dash_dash = 1; /* helps disambiguate */
}
arg = argv[0];
dash_dash_pos = -1;
for (i = 0; i < argc; i++) {
if (opts->accept_pathspec && !strcmp(argv[i], "--")) {
dash_dash_pos = i;
break;
}
}
if (dash_dash_pos == 0)
return 1; /* case (2) */
else if (dash_dash_pos == 1)
has_dash_dash = 1; /* case (3) or (1) */
else if (dash_dash_pos >= 2)
die(_("only one reference expected, %d given."), dash_dash_pos);
opts->count_checkout_paths = !opts->quiet && !has_dash_dash;
if (!strcmp(arg, "-"))
arg = "@{-1}";
if (repo_get_oid_mb(the_repository, arg, rev)) {
/*
* Either case (3) or (4), with <something> not being
* a commit, or an attempt to use case (1) with an
* invalid ref.
*
* It's likely an error, but we need to find out if
* we should auto-create the branch, case (3).(b).
*/
int recover_with_dwim = dwim_new_local_branch_ok;
checkout: disambiguate dwim tracking branches and local files When checkout dwim is added in [1], it is restricted to only dwim when certain conditions are met and fall back to default checkout behavior otherwise. It turns out falling back could be confusing. One of the conditions to turn git checkout frotz to git checkout -b frotz origin/frotz is that frotz must not exist as a file. But when the user comes to expect "git checkout frotz" to create the branch "frotz" and there happens to be a file named "frotz", git's silently reverting "frotz" file content is not helping. This is reported in Git mailing list [2] and even used as an example of "Git is bad" elsewhere [3]. We normally try to do the right thing, but when there are multiple "right things" to do, it's best to leave it to the user to decide. Check this case, ask the user to to disambiguate: - "git checkout -- foo" will check out path "foo" - "git checkout foo --" will dwim and create branch "foo" [4] For users who do not want dwim, use --no-guess. It's useless in this particular case because "git checkout --no-guess foo --" will just fail. But it could be used by scripts. [1] 70c9ac2f19 (DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz" - 2009-10-18) [2] https://public-inbox.org/git/CACsJy8B2TVr1g+k+eSQ=pBEO3WN4_LtgLo9gpur8X7Z9GOFL_A@mail.gmail.com/ [3] https://news.ycombinator.com/item?id=18230655 [4] a047fafc78 (checkout: allow dwim for branch creation for "git checkout $branch --" - 2013-10-18) Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-11-13 17:52:26 +00:00
int could_be_checkout_paths = !has_dash_dash &&
check_filename(opts->prefix, arg);
if (!has_dash_dash && !no_wildcard(arg))
recover_with_dwim = 0;
checkout: disambiguate dwim tracking branches and local files When checkout dwim is added in [1], it is restricted to only dwim when certain conditions are met and fall back to default checkout behavior otherwise. It turns out falling back could be confusing. One of the conditions to turn git checkout frotz to git checkout -b frotz origin/frotz is that frotz must not exist as a file. But when the user comes to expect "git checkout frotz" to create the branch "frotz" and there happens to be a file named "frotz", git's silently reverting "frotz" file content is not helping. This is reported in Git mailing list [2] and even used as an example of "Git is bad" elsewhere [3]. We normally try to do the right thing, but when there are multiple "right things" to do, it's best to leave it to the user to decide. Check this case, ask the user to to disambiguate: - "git checkout -- foo" will check out path "foo" - "git checkout foo --" will dwim and create branch "foo" [4] For users who do not want dwim, use --no-guess. It's useless in this particular case because "git checkout --no-guess foo --" will just fail. But it could be used by scripts. [1] 70c9ac2f19 (DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz" - 2009-10-18) [2] https://public-inbox.org/git/CACsJy8B2TVr1g+k+eSQ=pBEO3WN4_LtgLo9gpur8X7Z9GOFL_A@mail.gmail.com/ [3] https://news.ycombinator.com/item?id=18230655 [4] a047fafc78 (checkout: allow dwim for branch creation for "git checkout $branch --" - 2013-10-18) Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2018-11-13 17:52:26 +00:00
/*
* Accept "git checkout foo", "git checkout foo --"
* and "git switch foo" as candidates for dwim.
*/
if (!(argc == 1 && !has_dash_dash) &&
!(argc == 2 && has_dash_dash) &&
opts->accept_pathspec)
recover_with_dwim = 0;
if (recover_with_dwim) {
const char *remote = parse_remote_branch(arg, rev,
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
could_be_checkout_paths);
if (remote) {
*new_branch = arg;
arg = remote;
/* DWIMmed to create local branch, case (3).(b) */
} else {
recover_with_dwim = 0;
}
}
if (!recover_with_dwim) {
if (has_dash_dash)
die(_("invalid reference: %s"), arg);
return argcount;
}
}
/* we can't end up being in (2) anymore, eat the argument */
argcount++;
argv++;
argc--;
setup_new_branch_info_and_source_tree(new_branch_info, opts, rev, arg);
if (!opts->source_tree) /* case (1): want a tree */
die(_("reference is not a tree: %s"), arg);
if (!has_dash_dash) { /* case (3).(d) -> (1) */
/*
* Do not complain the most common case
* git checkout branch
* even if there happen to be a file called 'branch';
* it would be extremely annoying.
*/
if (argc)
verify_non_filename(opts->prefix, arg);
} else if (opts->accept_pathspec) {
argcount++;
argv++;
argc--;
}
return argcount;
}
static int switch_unborn_to_new_branch(const struct checkout_opts *opts)
{
int status;
struct strbuf branch_ref = STRBUF_INIT;
trace2_cmd_mode("unborn");
if (!opts->new_branch)
die(_("You are on a branch yet to be born"));
strbuf_addf(&branch_ref, "refs/heads/%s", opts->new_branch);
status = create_symref("HEAD", branch_ref.buf, "checkout -b");
strbuf_release(&branch_ref);
if (!opts->quiet)
fprintf(stderr, _("Switched to a new branch '%s'\n"),
opts->new_branch);
return status;
}
static void die_expecting_a_branch(const struct branch_info *branch_info)
{
struct object_id oid;
char *to_free;
int code;
if (repo_dwim_ref(the_repository, branch_info->name,
strlen(branch_info->name), &oid, &to_free, 0) == 1) {
const char *ref = to_free;
if (skip_prefix(ref, "refs/tags/", &ref))
code = die_message(_("a branch is expected, got tag '%s'"), ref);
else if (skip_prefix(ref, "refs/remotes/", &ref))
code = die_message(_("a branch is expected, got remote branch '%s'"), ref);
else
code = die_message(_("a branch is expected, got '%s'"), ref);
}
else if (branch_info->commit)
code = die_message(_("a branch is expected, got commit '%s'"), branch_info->name);
else
/*
* This case should never happen because we already die() on
* non-commit, but just in case.
*/
code = die_message(_("a branch is expected, got '%s'"), branch_info->name);
if (advice_enabled(ADVICE_SUGGEST_DETACHING_HEAD))
advise(_("If you want to detach HEAD at the commit, try again with the --detach option."));
exit(code);
}
static void die_if_some_operation_in_progress(void)
{
struct wt_status_state state;
memset(&state, 0, sizeof(state));
wt_status_get_state(the_repository, &state, 0);
if (state.merge_in_progress)
die(_("cannot switch branch while merging\n"
"Consider \"git merge --quit\" "
"or \"git worktree add\"."));
if (state.am_in_progress)
die(_("cannot switch branch in the middle of an am session\n"
"Consider \"git am --quit\" "
"or \"git worktree add\"."));
if (state.rebase_interactive_in_progress || state.rebase_in_progress)
die(_("cannot switch branch while rebasing\n"
"Consider \"git rebase --quit\" "
"or \"git worktree add\"."));
if (state.cherry_pick_in_progress)
die(_("cannot switch branch while cherry-picking\n"
"Consider \"git cherry-pick --quit\" "
"or \"git worktree add\"."));
if (state.revert_in_progress)
die(_("cannot switch branch while reverting\n"
"Consider \"git revert --quit\" "
"or \"git worktree add\"."));
if (state.bisect_in_progress)
warning(_("you are switching branch while bisecting"));
built-ins & libs & helpers: add/move destructors, fix leaks Fix various leaks in built-ins, libraries and a test helper here we were missing a call to strbuf_release(), string_list_clear() etc, or were calling them after a potential "return". Comments on individual changes: - builtin/checkout.c: Fix a memory leak that was introduced in [1]. A sibling leak introduced in [2] was recently fixed in [3]. As with [3] we should be using the wt_status_state_free_buffers() API introduced in [4]. - builtin/repack.c: Fix a leak that's been here since this use of "strbuf_release()" was added in a1bbc6c0176 (repack: rewrite the shell script in C, 2013-09-15). We don't use the variable for anything except this loop, so we can instead free it right afterwards. - builtin/rev-parse: Fix a leak that's been here since this code was added in 21d47835386 (Add a parseopt mode to git-rev-parse to bring parse-options to shell scripts., 2007-11-04). - builtin/stash.c: Fix a couple of leaks that have been here since this code was added in d4788af875c (stash: convert create to builtin, 2019-02-25), we strbuf_release()'d only some of the "struct strbuf" we allocated earlier in the function, let's release all of them. - ref-filter.c: Fix a leak in 482c1191869 (gpg-interface: improve interface for parsing tags, 2021-02-11), we don't use the "payload" variable that we ask parse_signature() to populate for us, so let's free it. - t/helper/test-fake-ssh.c: Fix a leak that's been here since this code was added in 3064d5a38c7 (mingw: fix t5601-clone.sh, 2016-01-27). Let's free the "struct strbuf" as soon as we don't need it anymore. 1. c45f0f525de (switch: reject if some operation is in progress, 2019-03-29) 2. 2708ce62d21 (branch: sort detached HEAD based on a flag, 2021-01-07) 3. abcac2e19fa (ref-filter.c: fix a leak in get_head_description, 2022-09-25) 4. 962dd7ebc3e (wt-status: introduce wt_status_state_free_buffers(), 2020-09-27). Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com> Signed-off-by: Taylor Blau <me@ttaylorr.com>
2022-11-08 18:17:42 +00:00
wt_status_state_free_buffers(&state);
}
/*
* die if attempting to checkout an existing branch that is in use
* in another worktree, unless ignore-other-wortrees option is given.
* The check is bypassed when the branch is already the current one,
* as it will not make things any worse.
*/
static void die_if_switching_to_a_branch_in_use(struct checkout_opts *opts,
const char *full_ref)
{
int flags;
char *head_ref;
if (opts->ignore_other_worktrees)
return;
head_ref = resolve_refdup("HEAD", 0, NULL, &flags);
if (head_ref && (!(flags & REF_ISSYMREF) || strcmp(head_ref, full_ref)))
die_if_checked_out(full_ref, 1);
free(head_ref);
}
static int checkout_branch(struct checkout_opts *opts,
struct branch_info *new_branch_info)
{
if (opts->pathspec.nr)
die(_("paths cannot be used with switching branches"));
if (opts->patch_mode)
die(_("'%s' cannot be used with switching branches"),
"--patch");
if (opts->overlay_mode != -1)
die(_("'%s' cannot be used with switching branches"),
"--[no]-overlay");
if (opts->writeout_stage)
die(_("'%s' cannot be used with switching branches"),
"--ours/--theirs");
if (opts->force && opts->merge)
die(_("'%s' cannot be used with '%s'"), "-f", "-m");
if (opts->discard_changes && opts->merge)
die(_("'%s' cannot be used with '%s'"), "--discard-changes", "--merge");
if (opts->force_detach && opts->new_branch)
die(_("'%s' cannot be used with '%s'"),
"--detach", "-b/-B/--orphan");
if (opts->new_orphan_branch) {
if (opts->track != BRANCH_TRACK_UNSPECIFIED)
die(_("'%s' cannot be used with '%s'"), "--orphan", "-t");
if (opts->orphan_from_empty_tree && new_branch_info->name)
die(_("'%s' cannot take <start-point>"), "--orphan");
} else if (opts->force_detach) {
if (opts->track != BRANCH_TRACK_UNSPECIFIED)
die(_("'%s' cannot be used with '%s'"), "--detach", "-t");
} else if (opts->track == BRANCH_TRACK_UNSPECIFIED)
opts->track = git_branch_track;
if (new_branch_info->name && !new_branch_info->commit)
die(_("Cannot switch branch to a non-commit '%s'"),
new_branch_info->name);
if (!opts->switch_branch_doing_nothing_is_ok &&
!new_branch_info->name &&
!opts->new_branch &&
!opts->force_detach)
die(_("missing branch or commit argument"));
if (!opts->implicit_detach &&
!opts->force_detach &&
!opts->new_branch &&
!opts->new_branch_force &&
new_branch_info->name &&
!new_branch_info->path)
die_expecting_a_branch(new_branch_info);
if (!opts->can_switch_when_in_progress)
die_if_some_operation_in_progress();
/* "git checkout <branch>" */
if (new_branch_info->path && !opts->force_detach && !opts->new_branch)
die_if_switching_to_a_branch_in_use(opts, new_branch_info->path);
/* "git checkout -B <branch>" */
if (opts->new_branch_force) {
char *full_ref = xstrfmt("refs/heads/%s", opts->new_branch);
die_if_switching_to_a_branch_in_use(opts, full_ref);
free(full_ref);
}
if (!new_branch_info->commit && opts->new_branch) {
struct object_id rev;
int flag;
if (!read_ref_full("HEAD", 0, &rev, &flag) &&
(flag & REF_ISSYMREF) && is_null_oid(&rev))
return switch_unborn_to_new_branch(opts);
}
return switch_branches(opts, new_branch_info);
}
static int parse_opt_conflict(const struct option *o, const char *arg, int unset)
{
struct checkout_opts *opts = o->value;
if (unset) {
opts->conflict_style = -1;
return 0;
}
opts->conflict_style = parse_conflict_style_name(arg);
if (opts->conflict_style < 0)
return error(_("unknown conflict style '%s'"), arg);
/* --conflict overrides a previous --no-merge */
if (!opts->merge)
opts->merge = -1;
return 0;
}
static struct option *add_common_options(struct checkout_opts *opts,
struct option *prevopts)
{
struct option options[] = {
OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
OPT_CALLBACK_F(0, "recurse-submodules", NULL,
"checkout", "control recursive updating of submodules",
PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater),
OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
OPT_CALLBACK(0, "conflict", opts, N_("style"),
N_("conflict style (merge, diff3, or zdiff3)"),
parse_opt_conflict),
OPT_END()
};
struct option *newopts = parse_options_concat(prevopts, options);
free(prevopts);
return newopts;
}
static struct option *add_common_switch_branch_options(
struct checkout_opts *opts, struct option *prevopts)
{
struct option options[] = {
OPT_BOOL('d', "detach", &opts->force_detach, N_("detach HEAD at named commit")),
OPT_CALLBACK_F('t', "track", &opts->track, "(direct|inherit)",
N_("set branch tracking configuration"),
PARSE_OPT_OPTARG,
parse_opt_tracking_mode),
OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
PARSE_OPT_NOCOMPLETE),
OPT_STRING(0, "orphan", &opts->new_orphan_branch, N_("new-branch"), N_("new unborn branch")),
OPT_BOOL_F(0, "overwrite-ignore", &opts->overwrite_ignore,
N_("update ignored files (default)"),
PARSE_OPT_NOCOMPLETE),
OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
N_("do not check if another worktree is holding the given ref")),
OPT_END()
};
struct option *newopts = parse_options_concat(prevopts, options);
free(prevopts);
return newopts;
}
static struct option *add_checkout_path_options(struct checkout_opts *opts,
struct option *prevopts)
{
struct option options[] = {
OPT_SET_INT_F('2', "ours", &opts->writeout_stage,
N_("checkout our version for unmerged files"),
2, PARSE_OPT_NONEG),
OPT_SET_INT_F('3', "theirs", &opts->writeout_stage,
N_("checkout their version for unmerged files"),
3, PARSE_OPT_NONEG),
OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
N_("do not limit pathspecs to sparse entries only")),
OPT_PATHSPEC_FROM_FILE(&opts->pathspec_from_file),
OPT_PATHSPEC_FILE_NUL(&opts->pathspec_file_nul),
OPT_END()
};
struct option *newopts = parse_options_concat(prevopts, options);
free(prevopts);
return newopts;
}
/* create-branch option (either b or c) */
static char cb_option = 'b';
static int checkout_main(int argc, const char **argv, const char *prefix,
struct checkout_opts *opts, struct option *options,
const char * const usagestr[])
{
int parseopt_flags = 0;
struct branch_info new_branch_info = { 0 };
int ret;
opts->overwrite_ignore = 1;
opts->prefix = prefix;
opts->show_progress = -1;
git_config(git_checkout_config, opts);
if (the_repository->gitdir) {
prepare_repo_settings(the_repository);
the_repository->settings.command_requires_full_index = 0;
}
checkout: stop expanding sparse indexes Previous changes did the necessary improvements to unpack-trees.c and diff-lib.c in order to modify a sparse index based on its comparision with a tree. The only remaining work is to remove some ensure_full_index() calls and add tests that verify that the index is not expanded in our interesting cases. Include 'switch' and 'restore' in these tests, as they share a base implementation with 'checkout'. Here are the relevant performance results from p2000-sparse-operations.sh: Test HEAD~1 HEAD -------------------------------------------------------------------------------- 2000.18: git checkout -f - (full-v3) 0.49(0.43+0.03) 0.47(0.39+0.05) -4.1% 2000.19: git checkout -f - (full-v4) 0.45(0.37+0.06) 0.42(0.37+0.05) -6.7% 2000.20: git checkout -f - (sparse-v3) 0.76(0.71+0.07) 0.04(0.03+0.04) -94.7% 2000.21: git checkout -f - (sparse-v4) 0.75(0.72+0.04) 0.05(0.06+0.04) -93.3% It is important to compare the full index case to the sparse index case, as the previous results for the sparse index were inflated by the index expansion. For index v4, this is an 88% improvement. On an internal repository with over two million paths at HEAD and a sparse-checkout definition containing ~60,000 of those paths, 'git checkout' went from 3.5s to 297ms with this change. The theoretical optimum where only those ~60,000 paths exist was 275ms, so the extra sparse directory entries contribute a 22ms overhead. Signed-off-by: Derrick Stolee <dstolee@microsoft.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2021-06-29 02:13:06 +00:00
opts->track = BRANCH_TRACK_UNSPECIFIED;
if (!opts->accept_pathspec && !opts->accept_ref)
BUG("make up your mind, you need to take _something_");
if (opts->accept_pathspec && opts->accept_ref)
parseopt_flags = PARSE_OPT_KEEP_DASHDASH;
argc = parse_options(argc, argv, prefix, options,
usagestr, parseopt_flags);
if (opts->show_progress < 0) {
if (opts->quiet)
opts->show_progress = 0;
else
opts->show_progress = isatty(2);
}
/* --conflicts implies --merge */
if (opts->merge == -1)
opts->merge = opts->conflict_style >= 0;
if (opts->force) {
opts->discard_changes = 1;
opts->ignore_unmerged_opt = "--force";
opts->ignore_unmerged = 1;
}
if ((!!opts->new_branch + !!opts->new_branch_force + !!opts->new_orphan_branch) > 1)
die(_("options '-%c', '-%c', and '%s' cannot be used together"),
cb_option, toupper(cb_option), "--orphan");
if (opts->overlay_mode == 1 && opts->patch_mode)
die(_("options '%s' and '%s' cannot be used together"), "-p", "--overlay");
if (opts->checkout_index >= 0 || opts->checkout_worktree >= 0) {
if (opts->checkout_index < 0)
opts->checkout_index = 0;
if (opts->checkout_worktree < 0)
opts->checkout_worktree = 0;
} else {
if (opts->checkout_index < 0)
opts->checkout_index = -opts->checkout_index - 1;
if (opts->checkout_worktree < 0)
opts->checkout_worktree = -opts->checkout_worktree - 1;
}
if (opts->checkout_index < 0 || opts->checkout_worktree < 0)
BUG("these flags should be non-negative by now");
/*
* convenient shortcut: "git restore --staged [--worktree]" equals
* "git restore --staged [--worktree] --source HEAD"
*/
if (!opts->from_treeish && opts->checkout_index)
opts->from_treeish = "HEAD";
/*
* From here on, new_branch will contain the branch to be checked out,
* and new_branch_force and new_orphan_branch will tell us which one of
* -b/-B/-c/-C/--orphan is being used.
*/
if (opts->new_branch_force)
opts->new_branch = opts->new_branch_force;
if (opts->new_orphan_branch)
opts->new_branch = opts->new_orphan_branch;
/* --track without -c/-C/-b/-B/--orphan should DWIM */
if (opts->track != BRANCH_TRACK_UNSPECIFIED && !opts->new_branch) {
const char *argv0 = argv[0];
if (!argc || !strcmp(argv0, "--"))
die(_("--track needs a branch name"));
skip_prefix(argv0, "refs/", &argv0);
skip_prefix(argv0, "remotes/", &argv0);
argv0 = strchr(argv0, '/');
if (!argv0 || !argv0[1])
die(_("missing branch name; try -%c"), cb_option);
opts->new_branch = argv0 + 1;
}
/*
* Extract branch name from command line arguments, so
* all that is left is pathspecs.
*
* Handle
*
* 1) git checkout <tree> -- [<paths>]
* 2) git checkout -- [<paths>]
* 3) git checkout <something> [<paths>]
*
* including "last branch" syntax and DWIM-ery for names of
* remote branches, erroring out for invalid or ambiguous cases.
*/
if (argc && opts->accept_ref) {
struct object_id rev;
int dwim_ok =
!opts->patch_mode &&
opts->dwim_new_local_branch &&
opts->track == BRANCH_TRACK_UNSPECIFIED &&
!opts->new_branch;
int n = parse_branchname_arg(argc, argv, dwim_ok,
&new_branch_info, opts, &rev);
argv += n;
argc -= n;
} else if (!opts->accept_ref && opts->from_treeish) {
struct object_id rev;
if (repo_get_oid_mb(the_repository, opts->from_treeish, &rev))
die(_("could not resolve %s"), opts->from_treeish);
setup_new_branch_info_and_source_tree(&new_branch_info,
opts, &rev,
opts->from_treeish);
if (!opts->source_tree)
die(_("reference is not a tree: %s"), opts->from_treeish);
}
if (argc) {
parse_pathspec(&opts->pathspec, 0,
opts->patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
prefix, argv);
if (!opts->pathspec.nr)
die(_("invalid path specification"));
/*
* Try to give more helpful suggestion.
* new_branch && argc > 1 will be caught later.
*/
if (opts->new_branch && argc == 1 && !new_branch_info.commit)
usability: don't ask questions if no reply is required There has been a bug report by a corporate user that stated that "spelling mistake of stash followed by a yes prints character 'y' infinite times." This analysis was false. When the spelling of a command contains errors, the git program tries to help the user by providing candidates which are close to the unexisting command. E.g Git prints the following: git: 'stahs' is not a git command. See 'git --help'. Did you mean this? stash and then exits. The problem with this hint is that it is not formally indicated as an hint and the user is in fact encouraged to reply to the question, whereas the Git command is already finished. The user was unlucky enough that it was the command he was looking for, and replied "yes" on the command line, effectively launching the `yes` program. The initial error is that the Git programs, when launched in command-line mode (without interaction) must not ask questions, because these questions would normally require a user input as a reply that they won't handle indeed. That's a source of confusion on UX level. To improve the general usability of the Git suite, the following rule was applied: if the sentence * appears in a non-interactive session * is printed last before exit * is a question addressing the user ("you") the sentence is turned into affirmative and proposes the option. The basic rewording of the question sentences has been extended to other spots found in the source. Requested at https://github.com/git/git-scm.com/issues/999 by rpai1 Signed-off-by: Jean-Noel Avila <jn.avila@free.fr> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-05-11 12:06:32 +00:00
die(_("'%s' is not a commit and a branch '%s' cannot be created from it"),
argv[0], opts->new_branch);
if (opts->force_detach)
die(_("git checkout: --detach does not take a path argument '%s'"),
argv[0]);
}
if (opts->pathspec_from_file) {
if (opts->pathspec.nr)
die(_("'%s' and pathspec arguments cannot be used together"), "--pathspec-from-file");
if (opts->force_detach)
die(_("options '%s' and '%s' cannot be used together"), "--pathspec-from-file", "--detach");
if (opts->patch_mode)
die(_("options '%s' and '%s' cannot be used together"), "--pathspec-from-file", "--patch");
parse_pathspec_file(&opts->pathspec, 0,
0,
prefix, opts->pathspec_from_file, opts->pathspec_file_nul);
} else if (opts->pathspec_file_nul) {
die(_("the option '%s' requires '%s'"), "--pathspec-file-nul", "--pathspec-from-file");
}
opts->pathspec.recursive = 1;
if (opts->pathspec.nr) {
if (1 < !!opts->writeout_stage + !!opts->force + !!opts->merge)
die(_("git checkout: --ours/--theirs, --force and --merge are incompatible when\n"
"checking out of the index."));
} else {
if (opts->accept_pathspec && !opts->empty_pathspec_ok &&
!opts->patch_mode) /* patch mode is special */
die(_("you must specify path(s) to restore"));
}
if (opts->new_branch) {
struct strbuf buf = STRBUF_INIT;
if (opts->new_branch_force)
opts->branch_exists = validate_branchname(opts->new_branch, &buf);
else
opts->branch_exists =
validate_new_branchname(opts->new_branch, &buf, 0);
strbuf_release(&buf);
}
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
if (opts->patch_mode || opts->pathspec.nr)
ret = checkout_paths(opts, &new_branch_info);
checkout: don't revert file on ambiguous tracking branches For easier understanding, here are the existing good scenarios: 1) Have *no* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will create local branch foo, see [1] and 1) Have *a* file 'foo', *no* local branch 'foo' and a *single* remote branch 'foo' 2) `git checkout foo` will complain, see [3] This patch prevents the following scenario: 1) Have *a* file 'foo', *no* local branch 'foo' and *multiple* remote branches 'foo' 2) `git checkout foo` will successfully... revert contents of file `foo`! That is, adding another remote suddenly changes behavior significantly, which is a surprise at best and could go unnoticed by user at worst. Please see [3] which gives some real world complaints. To my understanding, fix in [3] overlooked the case of multiple remotes, and the whole behavior of falling back to reverting file was never intended: [1] introduces the unexpected behavior. Before, there was fallback from not-a-ref to pathspec. This is reasonable fallback. After, there is another fallback from ambiguous-remote to pathspec. I understand that it was a copy&paste oversight. [2] noticed the unexpected behavior but chose to semi-document it instead of forbidding, because the goal of the patch series was focused on something else. [3] adds `die()` when there is ambiguity between branch and file. The case of multiple tracking branches is seemingly overlooked. The new behavior: if there is no local branch and multiple remote candidates, just die() and don't try reverting file whether it exists (prevents surprise) or not (improves error message). [1] Commit 70c9ac2f ("DWIM "git checkout frotz" to "git checkout -b frotz origin/frotz"" 2009-10-18) https://public-inbox.org/git/7vaazpxha4.fsf_-_@alter.siamese.dyndns.org/ [2] Commit ad8d5104 ("checkout: add advice for ambiguous "checkout <branch>"", 2018-06-05) https://public-inbox.org/git/20180502105452.17583-1-avarab@gmail.com/ [3] Commit be4908f1 ("checkout: disambiguate dwim tracking branches and local files", 2018-11-13) https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/ Signed-off-by: Alexandr Miloslavskiy <alexandr.miloslavskiy@syntevo.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-12-30 18:38:13 +00:00
else
ret = checkout_branch(opts, &new_branch_info);
branch_info_release(&new_branch_info);
clear_pathspec(&opts->pathspec);
free(opts->pathspec_from_file);
free(options);
return ret;
}
int cmd_checkout(int argc, const char **argv, const char *prefix)
{
struct checkout_opts opts = CHECKOUT_OPTS_INIT;
struct option *options;
struct option checkout_options[] = {
OPT_STRING('b', NULL, &opts.new_branch, N_("branch"),
N_("create and checkout a new branch")),
OPT_STRING('B', NULL, &opts.new_branch_force, N_("branch"),
N_("create/reset and checkout a branch")),
OPT_BOOL('l', NULL, &opts.new_branch_log, N_("create reflog for new branch")),
OPT_BOOL(0, "guess", &opts.dwim_new_local_branch,
N_("second guess 'git checkout <no-such-branch>' (default)")),
OPT_BOOL(0, "overlay", &opts.overlay_mode, N_("use overlay mode (default)")),
OPT_END()
};
opts.dwim_new_local_branch = 1;
opts.switch_branch_doing_nothing_is_ok = 1;
opts.only_merge_on_switching_branches = 0;
opts.accept_ref = 1;
opts.accept_pathspec = 1;
opts.implicit_detach = 1;
opts.can_switch_when_in_progress = 1;
opts.orphan_from_empty_tree = 0;
opts.empty_pathspec_ok = 1;
opts.overlay_mode = -1;
opts.checkout_index = -2; /* default on */
opts.checkout_worktree = -2; /* default on */
checkout: add simple check for 'git checkout -b' The 'git switch' command was created to separate half of the behavior of 'git checkout'. It specifically has the mode to do nothing with the index and working directory if the user only specifies to create a new branch and change HEAD to that branch. This is also the behavior most users expect from 'git checkout -b', but for historical reasons it also performs an index update by scanning the working directory. This can be slow for even moderately-sized repos. A performance fix for 'git checkout -b' was introduced by fa655d8411 (checkout: optimize "git checkout -b <new_branch>" 2018-08-16). That change includes details about the config setting checkout.optimizeNewBranch when the sparse-checkout feature is required. The way this change detected if this behavior change is safe was through the skip_merge_working_tree() method. This method was complex and needed to be updated as new options were introduced. This behavior was essentially reverted by 65f099b ("switch: no worktree status unless real branch switch happens" 2019-03-29). Instead, two members of the checkout_opts struct were used to distinguish between 'git checkout' and 'git switch': * switch_branch_doing_nothing_is_ok * only_merge_on_switching_branches These settings have opposite values depending on if we start in cmd_checkout or cmd_switch. The message for 64f099b includes "Users of big repos are encouraged to move to switch." Making this change while 'git switch' is still experimental is too aggressive. Create a happy medium between these two options by making 'git checkout -b <branch>' behave just like 'git switch', but only if we read exactly those arguments. This must be done in cmd_checkout to avoid the arguments being consumed by the option parsing logic. This differs from the previous change by fa644d8 in that the config option checkout.optimizeNewBranch remains deleted. This means that 'git checkout -b' will ignore the index merge even if we have a sparse-checkout file. While this is a behavior change for 'git checkout -b', it matches the behavior of 'git switch -c'. Signed-off-by: Derrick Stolee <dstolee@microsoft.com> Acked-by: Elijah Newren <newren@gmail.com> Acked-by: Taylor Blau <me@ttaylorr.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-08-29 17:01:32 +00:00
if (argc == 3 && !strcmp(argv[1], "-b")) {
/*
* User ran 'git checkout -b <branch>' and expects
* the same behavior as 'git switch -c <branch>'.
*/
opts.switch_branch_doing_nothing_is_ok = 0;
opts.only_merge_on_switching_branches = 1;
}
options = parse_options_dup(checkout_options);
options = add_common_options(&opts, options);
options = add_common_switch_branch_options(&opts, options);
options = add_checkout_path_options(&opts, options);
return checkout_main(argc, argv, prefix, &opts, options,
checkout_usage);
}
int cmd_switch(int argc, const char **argv, const char *prefix)
{
struct checkout_opts opts = CHECKOUT_OPTS_INIT;
struct option *options = NULL;
struct option switch_options[] = {
OPT_STRING('c', "create", &opts.new_branch, N_("branch"),
N_("create and switch to a new branch")),
OPT_STRING('C', "force-create", &opts.new_branch_force, N_("branch"),
N_("create/reset and switch to a branch")),
OPT_BOOL(0, "guess", &opts.dwim_new_local_branch,
N_("second guess 'git switch <no-such-branch>'")),
OPT_BOOL(0, "discard-changes", &opts.discard_changes,
N_("throw away local modifications")),
OPT_END()
};
opts.dwim_new_local_branch = 1;
opts.accept_ref = 1;
opts.accept_pathspec = 0;
opts.switch_branch_doing_nothing_is_ok = 0;
opts.only_merge_on_switching_branches = 1;
opts.implicit_detach = 0;
opts.can_switch_when_in_progress = 0;
opts.orphan_from_empty_tree = 1;
opts.overlay_mode = -1;
options = parse_options_dup(switch_options);
options = add_common_options(&opts, options);
options = add_common_switch_branch_options(&opts, options);
cb_option = 'c';
return checkout_main(argc, argv, prefix, &opts, options,
switch_branch_usage);
}
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
int cmd_restore(int argc, const char **argv, const char *prefix)
{
struct checkout_opts opts = CHECKOUT_OPTS_INIT;
struct option *options;
struct option restore_options[] = {
OPT_STRING('s', "source", &opts.from_treeish, "<tree-ish>",
N_("which tree-ish to checkout from")),
OPT_BOOL('S', "staged", &opts.checkout_index,
N_("restore the index")),
OPT_BOOL('W', "worktree", &opts.checkout_worktree,
N_("restore the working tree (default)")),
OPT_BOOL(0, "ignore-unmerged", &opts.ignore_unmerged,
N_("ignore unmerged entries")),
OPT_BOOL(0, "overlay", &opts.overlay_mode, N_("use overlay mode")),
OPT_END()
};
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
opts.accept_ref = 0;
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
opts.accept_pathspec = 1;
opts.empty_pathspec_ok = 0;
opts.overlay_mode = 0;
opts.checkout_index = -1; /* default off */
opts.checkout_worktree = -2; /* default on */
opts.ignore_unmerged_opt = "--ignore-unmerged";
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
options = parse_options_dup(restore_options);
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
options = add_common_options(&opts, options);
options = add_checkout_path_options(&opts, options);
return checkout_main(argc, argv, prefix, &opts, options,
restore_usage);
checkout: split part of it to new command 'restore' Previously the switching branch business of 'git checkout' becomes a new command 'switch'. This adds the restore command for the checking out paths path. Similar to git-switch, a new man page is added to describe what the command will become. The implementation will be updated shortly to match the man page. A couple main differences from 'git checkout <paths>': - 'restore' by default will only update worktree. This matters more when --source is specified ('checkout <tree> <paths>' updates both worktree and index). - 'restore --staged' can be used to restore the index. This command overlaps with 'git reset <paths>'. - both worktree and index could also be restored at the same time (from a tree) when both --staged and --worktree are specified. This overlaps with 'git checkout <tree> <paths>' - default source for restoring worktree and index is the index and HEAD respectively. A different (tree) source could be specified as with --source (*). - when both index and worktree are restored, --source must be specified since the default source for these two individual targets are different (**) - --no-overlay is enabled by default, if an entry is missing in the source, restoring means deleting the entry (*) I originally went with --from instead of --source. I still think --from is a better name. The short option -f however is already taken by force. And I do think short option is good to have, e.g. to write -s@ or -s@^ instead of --source=HEAD. (**) If you sit down and think about it, moving worktree's source from the index to HEAD makes sense, but nobody is really thinking it through when they type the commands. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
2019-04-25 09:45:45 +00:00
}