Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
#include "builtin.h"
|
2014-10-01 12:28:42 +02:00
|
|
|
#include "lockfile.h"
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
#include "parse-options.h"
|
|
|
|
#include "refs.h"
|
|
|
|
#include "commit.h"
|
|
|
|
#include "tree.h"
|
|
|
|
#include "tree-walk.h"
|
2009-04-20 12:58:18 +02:00
|
|
|
#include "cache-tree.h"
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
#include "unpack-trees.h"
|
|
|
|
#include "dir.h"
|
|
|
|
#include "run-command.h"
|
|
|
|
#include "merge-recursive.h"
|
|
|
|
#include "branch.h"
|
|
|
|
#include "diff.h"
|
|
|
|
#include "revision.h"
|
2008-02-17 02:17:09 +01:00
|
|
|
#include "remote.h"
|
2008-08-30 16:52:24 +02:00
|
|
|
#include "blob.h"
|
|
|
|
#include "xdiff-interface.h"
|
|
|
|
#include "ll-merge.h"
|
2009-12-25 09:30:51 +01:00
|
|
|
#include "resolve-undo.h"
|
2010-08-25 12:34:57 +02:00
|
|
|
#include "submodule.h"
|
2011-09-13 23:58:19 +02:00
|
|
|
#include "argv-array.h"
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
|
|
|
static const char * const checkout_usage[] = {
|
2012-08-20 14:31:59 +02:00
|
|
|
N_("git checkout [options] <branch>"),
|
|
|
|
N_("git checkout [options] [<branch>] -- <file>..."),
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
NULL,
|
|
|
|
};
|
|
|
|
|
2008-08-30 16:46:55 +02:00
|
|
|
struct checkout_opts {
|
2012-08-29 15:55:23 +02:00
|
|
|
int patch_mode;
|
2008-08-30 16:46:55 +02:00
|
|
|
int quiet;
|
|
|
|
int merge;
|
|
|
|
int force;
|
2011-02-08 11:32:49 +01:00
|
|
|
int force_detach;
|
2008-08-30 16:48:18 +02:00
|
|
|
int writeout_stage;
|
2011-11-27 11:15:33 +01:00
|
|
|
int overwrite_ignore;
|
2013-04-13 01:12:08 +02:00
|
|
|
int ignore_skipworktree;
|
2008-08-30 16:46:55 +02:00
|
|
|
|
|
|
|
const char *new_branch;
|
2010-06-23 21:29:00 +02:00
|
|
|
const char *new_branch_force;
|
2010-03-21 16:34:38 +01:00
|
|
|
const char *new_orphan_branch;
|
2008-08-30 16:46:55 +02:00
|
|
|
int new_branch_log;
|
|
|
|
enum branch_track track;
|
2010-08-25 12:34:57 +02:00
|
|
|
struct diff_options diff_options;
|
2012-08-29 15:55:23 +02:00
|
|
|
|
|
|
|
int branch_exists;
|
|
|
|
const char *prefix;
|
2013-07-14 10:35:41 +02:00
|
|
|
struct pathspec pathspec;
|
2012-08-29 15:55:23 +02:00
|
|
|
struct tree *source_tree;
|
2008-08-30 16:46:55 +02:00
|
|
|
};
|
|
|
|
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
static int post_checkout_hook(struct commit *old, struct commit *new,
|
|
|
|
int changed)
|
|
|
|
{
|
2014-03-18 11:00:53 +01:00
|
|
|
return run_hook_le(NULL, "post-checkout",
|
|
|
|
sha1_to_hex(old ? old->object.sha1 : null_sha1),
|
|
|
|
sha1_to_hex(new ? new->object.sha1 : null_sha1),
|
|
|
|
changed ? "1" : "0", NULL);
|
2009-01-16 20:09:58 +01:00
|
|
|
/* "new" can be NULL when checking out from the index before
|
|
|
|
a commit exists. */
|
2009-01-16 20:09:59 +01:00
|
|
|
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static int update_some(const unsigned char *sha1, const char *base, int baselen,
|
2008-07-14 21:22:12 +02:00
|
|
|
const char *pathname, unsigned mode, int stage, void *context)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
|
|
|
int len;
|
|
|
|
struct cache_entry *ce;
|
|
|
|
|
|
|
|
if (S_ISDIR(mode))
|
|
|
|
return READ_TREE_RECURSIVE;
|
|
|
|
|
|
|
|
len = baselen + strlen(pathname);
|
|
|
|
ce = xcalloc(1, cache_entry_size(len));
|
|
|
|
hashcpy(ce->sha1, sha1);
|
|
|
|
memcpy(ce->name, base, baselen);
|
|
|
|
memcpy(ce->name + baselen, pathname, len - baselen);
|
2012-07-11 11:22:37 +02:00
|
|
|
ce->ce_flags = create_ce_flags(0) | CE_UPDATE;
|
|
|
|
ce->ce_namelen = len;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
ce->ce_mode = create_ce_mode(mode);
|
|
|
|
add_cache_entry(ce, ADD_CACHE_OK_TO_ADD | ADD_CACHE_OK_TO_REPLACE);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-07-14 10:35:52 +02:00
|
|
|
static int read_tree_some(struct tree *tree, const struct pathspec *pathspec)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
2013-07-14 10:35:52 +02:00
|
|
|
read_tree_recursive(tree, "", 0, 0, pathspec, update_some, NULL);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
|
|
|
/* 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 17:29:00 +02:00
|
|
|
static int skip_same_name(const struct cache_entry *ce, int pos)
|
2008-08-29 22:40:36 +02:00
|
|
|
{
|
|
|
|
while (++pos < active_nr &&
|
|
|
|
!strcmp(active_cache[pos]->name, ce->name))
|
|
|
|
; /* skip */
|
|
|
|
return pos;
|
|
|
|
}
|
|
|
|
|
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 17:29:00 +02:00
|
|
|
static int check_stage(int stage, const struct cache_entry *ce, int pos)
|
2008-08-30 16:48:18 +02:00
|
|
|
{
|
|
|
|
while (pos < active_nr &&
|
|
|
|
!strcmp(active_cache[pos]->name, ce->name)) {
|
|
|
|
if (ce_stage(active_cache[pos]) == stage)
|
|
|
|
return 0;
|
|
|
|
pos++;
|
|
|
|
}
|
2011-02-23 00:41:40 +01:00
|
|
|
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);
|
2008-08-30 16:48:18 +02:00
|
|
|
}
|
|
|
|
|
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 17:29:00 +02:00
|
|
|
static int check_stages(unsigned stages, const struct cache_entry *ce, int pos)
|
2008-08-30 16:52:24 +02:00
|
|
|
{
|
2011-12-05 19:58:23 +01:00
|
|
|
unsigned seen = 0;
|
|
|
|
const char *name = ce->name;
|
|
|
|
|
|
|
|
while (pos < active_nr) {
|
|
|
|
ce = active_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);
|
2008-08-30 16:52:24 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-08-30 16:48:18 +02:00
|
|
|
static int checkout_stage(int stage, struct cache_entry *ce, int pos,
|
|
|
|
struct checkout *state)
|
|
|
|
{
|
|
|
|
while (pos < active_nr &&
|
|
|
|
!strcmp(active_cache[pos]->name, ce->name)) {
|
|
|
|
if (ce_stage(active_cache[pos]) == stage)
|
|
|
|
return checkout_entry(active_cache[pos], state, NULL);
|
|
|
|
pos++;
|
|
|
|
}
|
2011-02-23 00:41:40 +01:00
|
|
|
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);
|
2008-08-30 16:48:18 +02:00
|
|
|
}
|
2008-08-29 22:40:36 +02:00
|
|
|
|
2008-08-30 16:52:24 +02:00
|
|
|
static int checkout_merged(int pos, struct checkout *state)
|
|
|
|
{
|
|
|
|
struct cache_entry *ce = active_cache[pos];
|
|
|
|
const char *path = ce->name;
|
|
|
|
mmfile_t ancestor, ours, theirs;
|
|
|
|
int status;
|
|
|
|
unsigned char sha1[20];
|
|
|
|
mmbuffer_t result_buf;
|
2011-12-05 19:58:23 +01:00
|
|
|
unsigned char threeway[3][20];
|
2011-12-15 19:10:11 +01:00
|
|
|
unsigned mode = 0;
|
2008-08-30 16:52:24 +02:00
|
|
|
|
2011-12-05 19:58:23 +01:00
|
|
|
memset(threeway, 0, sizeof(threeway));
|
|
|
|
while (pos < active_nr) {
|
|
|
|
int stage;
|
|
|
|
stage = ce_stage(ce);
|
|
|
|
if (!stage || strcmp(path, ce->name))
|
|
|
|
break;
|
|
|
|
hashcpy(threeway[stage - 1], ce->sha1);
|
|
|
|
if (stage == 2)
|
|
|
|
mode = create_ce_mode(ce->ce_mode);
|
|
|
|
pos++;
|
|
|
|
ce = active_cache[pos];
|
|
|
|
}
|
|
|
|
if (is_null_sha1(threeway[1]) || is_null_sha1(threeway[2]))
|
|
|
|
return error(_("path '%s' does not have necessary versions"), path);
|
2008-08-30 16:52:24 +02:00
|
|
|
|
2011-12-05 19:58:23 +01:00
|
|
|
read_mmblob(&ancestor, threeway[0]);
|
|
|
|
read_mmblob(&ours, threeway[1]);
|
|
|
|
read_mmblob(&theirs, threeway[2]);
|
2008-08-30 16:52:24 +02:00
|
|
|
|
2010-08-05 13:24:58 +02:00
|
|
|
/*
|
|
|
|
* NEEDSWORK: re-create conflicts from merges with
|
|
|
|
* merge.renormalize set, too
|
|
|
|
*/
|
2010-03-21 01:40:19 +01:00
|
|
|
status = ll_merge(&result_buf, path, &ancestor, "base",
|
2010-08-26 07:49:53 +02:00
|
|
|
&ours, "ours", &theirs, "theirs", NULL);
|
2008-08-30 16:52:24 +02:00
|
|
|
free(ancestor.ptr);
|
|
|
|
free(ours.ptr);
|
|
|
|
free(theirs.ptr);
|
|
|
|
if (status < 0 || !result_buf.ptr) {
|
|
|
|
free(result_buf.ptr);
|
2011-02-23 00:41:39 +01:00
|
|
|
return error(_("path '%s': cannot merge"), path);
|
2008-08-30 16:52:24 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* NEEDSWORK:
|
|
|
|
* There is absolutely no reason to write this as a blob object
|
2009-04-17 20:13:30 +02:00
|
|
|
* and create a phony cache entry just to leak. This hack is
|
2008-08-30 16:52:24 +02:00
|
|
|
* 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_sha1_file(result_buf.ptr, result_buf.size,
|
|
|
|
blob_type, sha1))
|
2011-02-23 00:41:39 +01:00
|
|
|
die(_("Unable to add merge result for '%s'"), path);
|
2011-12-05 19:58:23 +01:00
|
|
|
ce = make_cache_entry(mode, sha1, path, 2, 0);
|
2008-10-05 04:14:40 +02:00
|
|
|
if (!ce)
|
2011-02-23 00:41:39 +01:00
|
|
|
die(_("make_cache_entry failed for path '%s'"), path);
|
2008-08-30 16:52:24 +02:00
|
|
|
status = checkout_entry(ce, state, NULL);
|
|
|
|
return status;
|
|
|
|
}
|
2008-08-29 22:40:36 +02:00
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
static int checkout_paths(const struct checkout_opts *opts,
|
|
|
|
const char *revision)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
|
|
|
int pos;
|
|
|
|
struct checkout state;
|
|
|
|
static char *ps_matched;
|
|
|
|
unsigned char rev[20];
|
|
|
|
int flag;
|
|
|
|
struct commit *head;
|
2008-05-28 23:48:57 +02:00
|
|
|
int errs = 0;
|
2012-08-30 14:45:50 +02:00
|
|
|
struct lock_file *lock_file;
|
|
|
|
|
|
|
|
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->force && opts->patch_mode)
|
|
|
|
die(_("'%s' cannot be used with updating paths"), "-f");
|
|
|
|
|
|
|
|
if (opts->force_detach)
|
|
|
|
die(_("'%s' cannot be used with updating paths"), "--detach");
|
|
|
|
|
|
|
|
if (opts->merge && opts->patch_mode)
|
|
|
|
die(_("'%s' cannot be used with %s"), "--merge", "--patch");
|
|
|
|
|
|
|
|
if (opts->force && opts->merge)
|
|
|
|
die(_("'%s' cannot be used with %s"), "-f", "-m");
|
|
|
|
|
|
|
|
if (opts->new_branch)
|
|
|
|
die(_("Cannot update paths and switch to branch '%s' at the same time."),
|
|
|
|
opts->new_branch);
|
|
|
|
|
|
|
|
if (opts->patch_mode)
|
|
|
|
return run_add_interactive(revision, "--patch=checkout",
|
2013-07-14 10:35:50 +02:00
|
|
|
&opts->pathspec);
|
2012-08-30 14:45:50 +02:00
|
|
|
|
|
|
|
lock_file = xcalloc(1, sizeof(struct lock_file));
|
2008-02-28 22:52:44 +01:00
|
|
|
|
2014-06-13 14:19:23 +02:00
|
|
|
hold_locked_index(lock_file, 1);
|
2013-07-14 10:35:49 +02:00
|
|
|
if (read_cache_preload(&opts->pathspec) < 0)
|
2011-02-23 00:41:39 +01:00
|
|
|
return error(_("corrupt index file"));
|
2008-02-28 22:52:44 +01:00
|
|
|
|
2012-08-29 15:55:23 +02:00
|
|
|
if (opts->source_tree)
|
2013-07-14 10:35:52 +02:00
|
|
|
read_tree_some(opts->source_tree, &opts->pathspec);
|
2008-02-28 22:52:44 +01:00
|
|
|
|
2013-07-14 10:35:41 +02:00
|
|
|
ps_matched = xcalloc(1, opts->pathspec.nr);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
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 06:58:21 +01:00
|
|
|
/*
|
|
|
|
* Make sure all pathspecs participated in locating the paths
|
|
|
|
* to be checked out.
|
|
|
|
*/
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
for (pos = 0; pos < active_nr; pos++) {
|
|
|
|
struct cache_entry *ce = active_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 06:58:21 +01:00
|
|
|
ce->ce_flags &= ~CE_MATCHED;
|
2013-04-13 01:12:08 +02:00
|
|
|
if (!opts->ignore_skipworktree && ce_skip_worktree(ce))
|
|
|
|
continue;
|
2012-08-29 15:55:23 +02:00
|
|
|
if (opts->source_tree && !(ce->ce_flags & CE_UPDATE))
|
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 06:58:21 +01:00
|
|
|
/*
|
|
|
|
* "git checkout tree-ish -- path", but this entry
|
|
|
|
* is in the original index; it will not be checked
|
|
|
|
* out to the working tree and it does not matter
|
|
|
|
* if pathspec matched this entry. We will not do
|
|
|
|
* anything to this entry at all.
|
|
|
|
*/
|
2011-09-30 19:33:15 +02:00
|
|
|
continue;
|
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 06:58:21 +01:00
|
|
|
/*
|
|
|
|
* 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_recursive (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.
|
|
|
|
*/
|
2014-01-24 14:40:28 +01:00
|
|
|
if (ce_path_match(ce, &opts->pathspec, ps_matched))
|
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 06:58:21 +01:00
|
|
|
ce->ce_flags |= CE_MATCHED;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
2013-07-14 10:35:53 +02:00
|
|
|
if (report_path_error(ps_matched, &opts->pathspec, opts->prefix)) {
|
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 06:58:21 +01:00
|
|
|
free(ps_matched);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
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 06:58:21 +01:00
|
|
|
}
|
|
|
|
free(ps_matched);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2009-12-25 20:57:11 +01:00
|
|
|
/* "checkout -m path" to recreate conflicted state */
|
|
|
|
if (opts->merge)
|
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 06:58:21 +01:00
|
|
|
unmerge_marked_index(&the_index);
|
2009-12-25 20:57:11 +01:00
|
|
|
|
2008-08-29 22:40:36 +02:00
|
|
|
/* Any unmerged paths? */
|
|
|
|
for (pos = 0; pos < active_nr; pos++) {
|
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 17:29:00 +02:00
|
|
|
const struct cache_entry *ce = active_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 06:58:21 +01:00
|
|
|
if (ce->ce_flags & CE_MATCHED) {
|
2008-08-29 22:40:36 +02:00
|
|
|
if (!ce_stage(ce))
|
|
|
|
continue;
|
2008-08-30 16:46:55 +02:00
|
|
|
if (opts->force) {
|
2011-02-23 00:41:39 +01:00
|
|
|
warning(_("path '%s' is unmerged"), ce->name);
|
2013-08-03 13:51:21 +02:00
|
|
|
} else if (opts->writeout_stage) {
|
|
|
|
errs |= check_stage(opts->writeout_stage, ce, pos);
|
2008-08-30 16:52:24 +02:00
|
|
|
} else if (opts->merge) {
|
2011-12-05 19:58:23 +01:00
|
|
|
errs |= check_stages((1<<2) | (1<<3), ce, pos);
|
2008-08-30 16:46:55 +02:00
|
|
|
} else {
|
|
|
|
errs = 1;
|
2011-02-23 00:41:39 +01:00
|
|
|
error(_("path '%s' is unmerged"), ce->name);
|
2008-08-30 16:46:55 +02:00
|
|
|
}
|
2008-08-29 22:40:36 +02:00
|
|
|
pos = skip_same_name(ce, pos) - 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (errs)
|
|
|
|
return 1;
|
|
|
|
|
2008-05-28 23:48:57 +02:00
|
|
|
/* Now we are committed to check them out */
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
memset(&state, 0, sizeof(state));
|
|
|
|
state.force = 1;
|
|
|
|
state.refresh_cache = 1;
|
2014-06-13 14:19:34 +02:00
|
|
|
state.istate = &the_index;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
for (pos = 0; pos < active_nr; pos++) {
|
|
|
|
struct cache_entry *ce = active_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 06:58:21 +01:00
|
|
|
if (ce->ce_flags & CE_MATCHED) {
|
2008-08-29 22:40:36 +02:00
|
|
|
if (!ce_stage(ce)) {
|
|
|
|
errs |= checkout_entry(ce, &state, NULL);
|
|
|
|
continue;
|
|
|
|
}
|
2013-08-03 13:51:21 +02:00
|
|
|
if (opts->writeout_stage)
|
|
|
|
errs |= checkout_stage(opts->writeout_stage, ce, pos, &state);
|
|
|
|
else if (opts->merge)
|
2008-08-30 16:52:24 +02:00
|
|
|
errs |= checkout_merged(pos, &state);
|
2008-08-29 22:40:36 +02:00
|
|
|
pos = skip_same_name(ce, pos) - 1;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-06-13 14:19:23 +02:00
|
|
|
if (write_locked_index(&the_index, lock_file, COMMIT_LOCK))
|
2011-02-23 00:41:39 +01:00
|
|
|
die(_("unable to write new index file"));
|
2008-02-28 22:52:44 +01:00
|
|
|
|
2014-07-15 21:59:36 +02:00
|
|
|
read_ref_full("HEAD", 0, rev, &flag);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
head = lookup_commit_reference_gently(rev, 1);
|
|
|
|
|
2008-05-28 23:48:57 +02:00
|
|
|
errs |= post_checkout_hook(head, head, 0);
|
|
|
|
return errs;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
2012-08-29 15:55:22 +02:00
|
|
|
static void show_local_changes(struct object *head,
|
|
|
|
const struct diff_options *opts)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
|
|
|
struct rev_info rev;
|
|
|
|
/* I think we want full paths, even if we're in a subdirectory. */
|
|
|
|
init_revisions(&rev, NULL);
|
2010-08-25 12:34:57 +02:00
|
|
|
rev.diffopt.flags = opts->flags;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
rev.diffopt.output_format |= DIFF_FORMAT_NAME_STATUS;
|
2012-08-03 14:16:24 +02:00
|
|
|
diff_setup_done(&rev.diffopt);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
add_pending_object(&rev, head, NULL);
|
|
|
|
run_diff_index(&rev, 0);
|
|
|
|
}
|
|
|
|
|
2011-02-22 23:43:22 +01:00
|
|
|
static void describe_detached_head(const char *msg, struct commit *commit)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
2008-10-09 21:12:12 +02:00
|
|
|
struct strbuf sb = STRBUF_INIT;
|
2013-10-24 10:54:53 +02:00
|
|
|
if (!parse_commit(commit))
|
|
|
|
pp_commit_easy(CMIT_FMT_ONELINE, commit, &sb);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
fprintf(stderr, "%s %s... %s\n", msg,
|
|
|
|
find_unique_abbrev(commit->object.sha1, DEFAULT_ABBREV), sb.buf);
|
|
|
|
strbuf_release(&sb);
|
|
|
|
}
|
|
|
|
|
2012-08-29 15:55:22 +02:00
|
|
|
static int reset_tree(struct tree *tree, const struct checkout_opts *o,
|
|
|
|
int worktree, int *writeout_error)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
|
|
|
struct unpack_trees_options opts;
|
|
|
|
struct tree_desc tree_desc;
|
2008-03-06 21:26:14 +01:00
|
|
|
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
memset(&opts, 0, sizeof(opts));
|
|
|
|
opts.head_idx = -1;
|
2008-05-28 23:59:40 +02:00
|
|
|
opts.update = worktree;
|
|
|
|
opts.skip_unmerged = !worktree;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
opts.reset = 1;
|
|
|
|
opts.merge = 1;
|
|
|
|
opts.fn = oneway_merge;
|
2012-05-24 08:12:24 +02:00
|
|
|
opts.verbose_update = !o->quiet && isatty(2);
|
2008-03-07 03:12:28 +01:00
|
|
|
opts.src_index = &the_index;
|
|
|
|
opts.dst_index = &the_index;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
parse_tree(tree);
|
|
|
|
init_tree_desc(&tree_desc, tree->buffer, tree->size);
|
2008-05-29 00:26:59 +02:00
|
|
|
switch (unpack_trees(1, &tree_desc, &opts)) {
|
|
|
|
case -2:
|
2012-08-29 15:55:22 +02:00
|
|
|
*writeout_error = 1;
|
2008-05-29 00:26:59 +02:00
|
|
|
/*
|
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
case 0:
|
|
|
|
return 0;
|
|
|
|
default:
|
2008-05-28 23:54:02 +02:00
|
|
|
return 128;
|
2008-05-29 00:26:59 +02:00
|
|
|
}
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
struct branch_info {
|
|
|
|
const char *name; /* The short name used */
|
|
|
|
const char *path; /* The full name of a real branch */
|
|
|
|
struct commit *commit; /* The named commit */
|
|
|
|
};
|
|
|
|
|
|
|
|
static void setup_branch_path(struct branch_info *branch)
|
|
|
|
{
|
2008-10-09 21:12:12 +02:00
|
|
|
struct strbuf buf = STRBUF_INIT;
|
2009-01-17 17:09:53 +01:00
|
|
|
|
2009-03-21 21:17:30 +01:00
|
|
|
strbuf_branchname(&buf, branch->name);
|
|
|
|
if (strcmp(buf.buf, branch->name))
|
2009-01-17 17:09:53 +01:00
|
|
|
branch->name = xstrdup(buf.buf);
|
2009-03-21 21:17:30 +01:00
|
|
|
strbuf_splice(&buf, 0, 0, "refs/heads/", 11);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
branch->path = strbuf_detach(&buf, NULL);
|
|
|
|
}
|
|
|
|
|
2012-08-29 15:55:22 +02:00
|
|
|
static int merge_working_tree(const struct checkout_opts *opts,
|
|
|
|
struct branch_info *old,
|
|
|
|
struct branch_info *new,
|
|
|
|
int *writeout_error)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
struct lock_file *lock_file = xcalloc(1, sizeof(struct lock_file));
|
2008-12-06 02:54:10 +01:00
|
|
|
|
2014-06-13 14:19:23 +02:00
|
|
|
hold_locked_index(lock_file, 1);
|
2009-05-10 00:11:17 +02:00
|
|
|
if (read_cache_preload(NULL) < 0)
|
2011-02-23 00:41:39 +01:00
|
|
|
return error(_("corrupt index file"));
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2009-12-25 09:30:51 +01:00
|
|
|
resolve_undo_clear();
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
if (opts->force) {
|
2012-08-29 15:55:22 +02:00
|
|
|
ret = reset_tree(new->commit->tree, opts, 1, writeout_error);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
} else {
|
|
|
|
struct tree_desc trees[2];
|
|
|
|
struct tree *tree;
|
|
|
|
struct unpack_trees_options topts;
|
2008-03-06 21:26:14 +01:00
|
|
|
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
memset(&topts, 0, sizeof(topts));
|
|
|
|
topts.head_idx = -1;
|
2008-03-07 03:12:28 +01:00
|
|
|
topts.src_index = &the_index;
|
|
|
|
topts.dst_index = &the_index;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2010-09-02 13:57:34 +02:00
|
|
|
setup_unpack_trees_porcelain(&topts, "checkout");
|
2008-05-17 21:03:49 +02:00
|
|
|
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
refresh_cache(REFRESH_QUIET);
|
|
|
|
|
|
|
|
if (unmerged_cache()) {
|
2011-02-23 00:41:39 +01:00
|
|
|
error(_("you need to resolve your current index first"));
|
2008-02-24 00:45:19 +01:00
|
|
|
return 1;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
2008-02-24 00:45:19 +01:00
|
|
|
|
|
|
|
/* 2-way merge to the new branch */
|
checkout: Fix "initial checkout" detection
Earlier commit 5521883 (checkout: do not lose staged removal, 2008-09-07)
tightened the rule to prevent switching branches from losing local
changes, so that staged removal of paths can be protected, while
attempting to keep a loophole to still allow a special case of switching
out of an un-checked-out state.
However, the loophole was made a bit too tight, and did not allow
switching from one branch (in an un-checked-out state) to check out
another branch.
The change to builtin-checkout.c in this commit loosens it to allow this,
by not insisting the original commit and the new commit to be the same.
It also introduces a new function, is_index_unborn (and an associated
macro, is_cache_unborn), to check if the repository is truly in an
un-checked-out state more reliably, by making sure that $GIT_INDEX_FILE
did not exist when populating the in-core index structure. A few places
the earlier commit 5521883 added the check for the initial checkout
condition are updated to use this function.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-11-12 20:52:35 +01:00
|
|
|
topts.initial_checkout = is_cache_unborn();
|
2008-02-24 00:45:19 +01:00
|
|
|
topts.update = 1;
|
|
|
|
topts.merge = 1;
|
2010-01-07 08:51:47 +01:00
|
|
|
topts.gently = opts->merge && old->commit;
|
2012-05-24 08:12:24 +02:00
|
|
|
topts.verbose_update = !opts->quiet && isatty(2);
|
2008-02-24 00:45:19 +01:00
|
|
|
topts.fn = twoway_merge;
|
2011-11-27 11:15:33 +01:00
|
|
|
if (opts->overwrite_ignore) {
|
|
|
|
topts.dir = xcalloc(1, sizeof(*topts.dir));
|
|
|
|
topts.dir->flags |= DIR_SHOW_IGNORED;
|
|
|
|
setup_standard_excludes(topts.dir);
|
|
|
|
}
|
checkout: do not imply "-f" on unborn branches
When checkout sees that HEAD points to a non-existent ref,
it currently acts as if "-f" was given; this behavior dates
back to 5a03e7f, which enabled checkout from unborn branches
in the shell version of "git-checkout". The reasoning given
is to avoid the code path which tries to merge the tree
contents. When checkout was converted to C, this code
remained intact.
The unfortunate side effect of this strategy is that the
"force" code path will overwrite working tree and index
state that may be precious to the user. Instead of enabling
"force", this patch uses the normal "merge" codepath for an
unborn branch, but substitutes the empty tree for the "old"
commit.
This means that in the absence of an index, any files in the
working tree will be treated as untracked files, and a
checkout which would overwrite them is aborted. Similarly,
any paths in the index will be merged with an empty entry
as the base, meaning that unless the new branch's content is
identical to what's in the index, there will be a conflict
and the checkout will be aborted.
The user is then free to correct the situation or proceed
with "-f" as appropriate.
This patch also removes the "warning: you are on a branch
yet to be born" message. Its function was to warn the user
that we were enabling the "-f" option. Since we are no
longer doing that, there is no reason for the user to care
whether we are switching away from an unborn branch.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-08-25 05:03:16 +02:00
|
|
|
tree = parse_tree_indirect(old->commit ?
|
|
|
|
old->commit->object.sha1 :
|
2011-02-07 09:17:27 +01:00
|
|
|
EMPTY_TREE_SHA1_BIN);
|
2008-02-24 00:45:19 +01:00
|
|
|
init_tree_desc(&trees[0], tree->buffer, tree->size);
|
|
|
|
tree = parse_tree_indirect(new->commit->object.sha1);
|
|
|
|
init_tree_desc(&trees[1], tree->buffer, tree->size);
|
|
|
|
|
2008-05-29 00:26:59 +02:00
|
|
|
ret = unpack_trees(2, trees, &topts);
|
2009-05-12 18:41:28 +02:00
|
|
|
if (ret == -1) {
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
/*
|
|
|
|
* 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 *result;
|
|
|
|
struct tree *work;
|
2008-08-25 16:25:57 +02:00
|
|
|
struct merge_options o;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
if (!opts->merge)
|
|
|
|
return 1;
|
2010-01-07 08:51:47 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Without old->commit, the below is the same as
|
|
|
|
* the two-tree unpack we already tried and failed.
|
|
|
|
*/
|
|
|
|
if (!old->commit)
|
|
|
|
return 1;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
|
|
|
/* 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.
|
|
|
|
*/
|
|
|
|
|
2008-05-12 19:58:10 +02:00
|
|
|
add_files_to_cache(NULL, NULL, 0);
|
2010-08-05 13:32:41 +02:00
|
|
|
/*
|
|
|
|
* NEEDSWORK: carrying over local changes
|
|
|
|
* when branches have different end-of-line
|
|
|
|
* normalization (or clean+smudge rules) is
|
|
|
|
* a pain; plumb in an option to set
|
|
|
|
* o.renormalize?
|
|
|
|
*/
|
2008-08-25 16:25:57 +02:00
|
|
|
init_merge_options(&o);
|
|
|
|
o.verbosity = 0;
|
|
|
|
work = write_tree_from_memory(&o);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2012-08-29 15:55:22 +02:00
|
|
|
ret = reset_tree(new->commit->tree, opts, 1,
|
|
|
|
writeout_error);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
2010-03-21 01:42:51 +01:00
|
|
|
o.ancestor = old->name;
|
2008-08-25 16:25:57 +02:00
|
|
|
o.branch1 = new->name;
|
|
|
|
o.branch2 = "local";
|
|
|
|
merge_trees(&o, new->commit->tree, work,
|
|
|
|
old->commit->tree, &result);
|
2012-08-29 15:55:22 +02:00
|
|
|
ret = reset_tree(new->commit->tree, opts, 0,
|
|
|
|
writeout_error);
|
2008-05-28 23:54:02 +02:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-07-06 06:06:56 +02:00
|
|
|
if (!active_cache_tree)
|
|
|
|
active_cache_tree = cache_tree();
|
|
|
|
|
|
|
|
if (!cache_tree_fully_valid(active_cache_tree))
|
2014-09-11 19:33:32 +02:00
|
|
|
cache_tree_update(&the_index, WRITE_TREE_SILENT | WRITE_TREE_REPAIR);
|
2014-07-06 06:06:56 +02:00
|
|
|
|
2014-06-13 14:19:23 +02:00
|
|
|
if (write_locked_index(&the_index, lock_file, COMMIT_LOCK))
|
2011-02-23 00:41:39 +01:00
|
|
|
die(_("unable to write new index file"));
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2008-09-25 10:35:38 +02:00
|
|
|
if (!opts->force && !opts->quiet)
|
2010-08-25 12:34:57 +02:00
|
|
|
show_local_changes(&new->commit->object, &opts->diff_options);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-07-02 09:51:18 +02:00
|
|
|
static void report_tracking(struct branch_info *new)
|
2008-02-17 02:17:09 +01:00
|
|
|
{
|
2008-07-02 09:51:18 +02:00
|
|
|
struct strbuf sb = STRBUF_INIT;
|
2008-02-21 04:42:53 +01:00
|
|
|
struct branch *branch = branch_get(new->name);
|
2008-02-17 02:17:09 +01:00
|
|
|
|
2008-07-02 09:51:18 +02:00
|
|
|
if (!format_tracking_info(branch, &sb))
|
2008-02-17 02:17:09 +01:00
|
|
|
return;
|
2008-07-02 09:51:18 +02:00
|
|
|
fputs(sb.buf, stdout);
|
|
|
|
strbuf_release(&sb);
|
2008-02-21 00:05:23 +01:00
|
|
|
}
|
2008-02-17 02:17:09 +01:00
|
|
|
|
2012-08-29 15:55:22 +02:00
|
|
|
static void update_refs_for_switch(const struct checkout_opts *opts,
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
struct branch_info *old,
|
|
|
|
struct branch_info *new)
|
|
|
|
{
|
2008-10-09 21:12:12 +02:00
|
|
|
struct strbuf msg = STRBUF_INIT;
|
2013-06-16 10:45:16 +02:00
|
|
|
const char *old_desc, *reflog_msg;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
if (opts->new_branch) {
|
2010-05-22 02:28:37 +02:00
|
|
|
if (opts->new_orphan_branch) {
|
|
|
|
if (opts->new_branch_log && !log_all_ref_updates) {
|
|
|
|
int temp;
|
2014-11-30 09:24:28 +01:00
|
|
|
struct strbuf log_file = STRBUF_INIT;
|
|
|
|
int ret;
|
|
|
|
const char *ref_name;
|
2010-05-22 02:28:37 +02:00
|
|
|
|
2014-11-30 09:24:28 +01:00
|
|
|
ref_name = mkpath("refs/heads/%s", opts->new_orphan_branch);
|
2010-05-22 02:28:37 +02:00
|
|
|
temp = log_all_ref_updates;
|
|
|
|
log_all_ref_updates = 1;
|
2014-11-30 09:24:28 +01:00
|
|
|
ret = log_ref_setup(ref_name, &log_file);
|
|
|
|
log_all_ref_updates = temp;
|
|
|
|
strbuf_release(&log_file);
|
|
|
|
if (ret) {
|
2011-02-23 00:41:39 +01:00
|
|
|
fprintf(stderr, _("Can not do reflog for '%s'\n"),
|
2010-05-22 02:28:37 +02:00
|
|
|
opts->new_orphan_branch);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
2010-06-23 21:29:00 +02:00
|
|
|
create_branch(old->name, opts->new_branch, new->name,
|
|
|
|
opts->new_branch_force ? 1 : 0,
|
2011-11-26 09:54:55 +01:00
|
|
|
opts->new_branch_log,
|
|
|
|
opts->new_branch_force ? 1 : 0,
|
2012-03-27 01:51:01 +02:00
|
|
|
opts->quiet,
|
2011-11-26 09:54:55 +01:00
|
|
|
opts->track);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
new->name = opts->new_branch;
|
|
|
|
setup_branch_path(new);
|
|
|
|
}
|
|
|
|
|
|
|
|
old_desc = old->name;
|
2008-11-08 13:03:59 +01:00
|
|
|
if (!old_desc && old->commit)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
old_desc = sha1_to_hex(old->commit->object.sha1);
|
2013-06-16 10:45:16 +02:00
|
|
|
|
|
|
|
reflog_msg = getenv("GIT_REFLOG_ACTION");
|
|
|
|
if (!reflog_msg)
|
|
|
|
strbuf_addf(&msg, "checkout: moving from %s to %s",
|
|
|
|
old_desc ? old_desc : "(invalid)", new->name);
|
|
|
|
else
|
|
|
|
strbuf_insert(&msg, 0, reflog_msg, strlen(reflog_msg));
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2011-02-08 11:34:34 +01:00
|
|
|
if (!strcmp(new->name, "HEAD") && !new->path && !opts->force_detach) {
|
|
|
|
/* Nothing to do. */
|
|
|
|
} else if (opts->force_detach || !new->path) { /* No longer on any branch. */
|
|
|
|
update_ref(msg.buf, "HEAD", new->commit->object.sha1, NULL,
|
2014-04-07 15:47:56 +02:00
|
|
|
REF_NODEREF, UPDATE_REFS_DIE_ON_ERR);
|
2011-02-08 11:34:34 +01:00
|
|
|
if (!opts->quiet) {
|
|
|
|
if (old->path && advice_detached_head)
|
2012-01-16 10:46:16 +01:00
|
|
|
detach_advice(new->name);
|
2011-04-02 02:55:55 +02:00
|
|
|
describe_detached_head(_("HEAD is now at"), new->commit);
|
2011-02-08 11:34:34 +01:00
|
|
|
}
|
|
|
|
} else if (new->path) { /* Switch branches. */
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
create_symref("HEAD", new->path, msg.buf);
|
|
|
|
if (!opts->quiet) {
|
2011-02-23 00:41:43 +01:00
|
|
|
if (old->path && !strcmp(new->path, old->path)) {
|
2011-11-26 09:54:55 +01:00
|
|
|
if (opts->new_branch_force)
|
|
|
|
fprintf(stderr, _("Reset branch '%s'\n"),
|
|
|
|
new->name);
|
|
|
|
else
|
|
|
|
fprintf(stderr, _("Already on '%s'\n"),
|
|
|
|
new->name);
|
2011-02-23 00:41:43 +01:00
|
|
|
} else if (opts->new_branch) {
|
|
|
|
if (opts->branch_exists)
|
|
|
|
fprintf(stderr, _("Switched to and reset branch '%s'\n"), new->name);
|
|
|
|
else
|
|
|
|
fprintf(stderr, _("Switched to a new branch '%s'\n"), new->name);
|
|
|
|
} else {
|
2011-02-23 00:41:39 +01:00
|
|
|
fprintf(stderr, _("Switched to branch '%s'\n"),
|
2010-08-20 19:41:47 +02:00
|
|
|
new->name);
|
2011-02-23 00:41:43 +01:00
|
|
|
}
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
2010-05-22 02:28:37 +02:00
|
|
|
if (old->path && old->name) {
|
2014-05-07 00:45:53 +02:00
|
|
|
if (!ref_exists(old->path) && reflog_exists(old->path))
|
2014-05-07 00:45:52 +02:00
|
|
|
delete_reflog(old->path);
|
2010-05-22 02:28:37 +02:00
|
|
|
}
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
remove_branch_state();
|
|
|
|
strbuf_release(&msg);
|
2011-02-08 11:32:49 +01:00
|
|
|
if (!opts->quiet &&
|
|
|
|
(new->path || (!opts->force_detach && !strcmp(new->name, "HEAD"))))
|
2008-07-02 09:51:18 +02:00
|
|
|
report_tracking(new);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
2011-10-01 17:51:39 +02:00
|
|
|
static int add_pending_uninteresting_ref(const char *refname,
|
|
|
|
const unsigned char *sha1,
|
|
|
|
int flags, void *cb_data)
|
2011-02-19 01:04:47 +01:00
|
|
|
{
|
checkout: don't confuse ref and object flags
When we are leaving a detached HEAD, we do a revision traversal to
check whether we are orphaning any commits, marking the commit we're
leaving as the start of the traversal, and all existing refs as
uninteresting.
Prior to commit 468224e5, we did so by calling for_each_ref, and
feeding each resulting refname to setup_revisions. Commit 468224e5
refactored this to simply mark the pending objects, saving an extra
lookup.
However, it confused the "flags" parameter to the each_ref_fn
clalback, which is about the flags we found while looking up the ref
with the object flag. Because REF_ISSYMREF ("this ref is a symbolic
ref, e.g. refs/remotes/origin/HEAD") happens to be the same bit
pattern as SEEN ("we have picked this object up from the pending
list and moved it to revs.commits list"), we incorrectly reported
that a commit previously at the detached HEAD will become
unreachable if the only ref that can reach the commit happens to be
pointed at by a symbolic ref.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2012-07-25 23:57:30 +02:00
|
|
|
add_pending_sha1(cb_data, refname, sha1, UNINTERESTING);
|
2011-03-20 10:09:18 +01:00
|
|
|
return 0;
|
|
|
|
}
|
2011-02-19 01:04:47 +01:00
|
|
|
|
|
|
|
static void describe_one_orphan(struct strbuf *sb, struct commit *commit)
|
|
|
|
{
|
2011-03-20 10:19:01 +01:00
|
|
|
strbuf_addstr(sb, " ");
|
|
|
|
strbuf_addstr(sb,
|
|
|
|
find_unique_abbrev(commit->object.sha1, DEFAULT_ABBREV));
|
|
|
|
strbuf_addch(sb, ' ');
|
2013-10-24 10:54:53 +02:00
|
|
|
if (!parse_commit(commit))
|
|
|
|
pp_commit_easy(CMIT_FMT_ONELINE, commit, sb);
|
2011-02-19 01:04:47 +01:00
|
|
|
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
|
2011-04-10 21:34:08 +02:00
|
|
|
strbuf_addf(&sb, _(" ... and %d more.\n"), more);
|
2011-02-19 01:04:47 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stderr,
|
2011-04-10 21:34:08 +02:00
|
|
|
Q_(
|
|
|
|
/* The singular version */
|
|
|
|
"Warning: you are leaving %d commit behind, "
|
|
|
|
"not connected to\n"
|
|
|
|
"any of your branches:\n\n"
|
2011-06-30 02:03:10 +02:00
|
|
|
"%s\n",
|
2011-04-10 21:34:08 +02:00
|
|
|
/* The plural version */
|
|
|
|
"Warning: you are leaving %d commits behind, "
|
2011-02-19 01:04:47 +01:00
|
|
|
"not connected to\n"
|
|
|
|
"any of your branches:\n\n"
|
2011-05-24 19:08:29 +02:00
|
|
|
"%s\n",
|
2011-04-10 21:34:08 +02:00
|
|
|
/* Give ngettext() the count */
|
|
|
|
lost),
|
|
|
|
lost,
|
2011-05-24 19:08:29 +02:00
|
|
|
sb.buf);
|
2011-02-19 01:04:47 +01:00
|
|
|
strbuf_release(&sb);
|
2011-05-24 19:08:29 +02:00
|
|
|
|
|
|
|
if (advice_detached_head)
|
|
|
|
fprintf(stderr,
|
2011-06-30 02:03:10 +02:00
|
|
|
_(
|
2011-05-24 19:08:29 +02:00
|
|
|
"If you want to keep them by creating a new branch, "
|
|
|
|
"this may be a good time\nto do so with:\n\n"
|
2011-06-30 02:03:10 +02:00
|
|
|
" git branch new_branch_name %s\n\n"),
|
2013-04-08 17:53:42 +02:00
|
|
|
find_unique_abbrev(commit->object.sha1, DEFAULT_ABBREV));
|
2011-02-19 01:04:47 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 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.
|
|
|
|
*/
|
2012-05-04 20:14:48 +02:00
|
|
|
static void orphaned_commit_warning(struct commit *old, struct commit *new)
|
2011-02-19 01:04:47 +01:00
|
|
|
{
|
|
|
|
struct rev_info revs;
|
2012-05-04 20:14:48 +02:00
|
|
|
struct object *object = &old->object;
|
2011-10-01 18:09:36 +02:00
|
|
|
struct object_array refs;
|
2011-02-19 01:04:47 +01:00
|
|
|
|
|
|
|
init_revisions(&revs, NULL);
|
2011-10-01 17:51:39 +02:00
|
|
|
setup_revisions(0, NULL, &revs, NULL);
|
|
|
|
|
|
|
|
object->flags &= ~UNINTERESTING;
|
|
|
|
add_pending_object(&revs, object, sha1_to_hex(object->sha1));
|
|
|
|
|
|
|
|
for_each_ref(add_pending_uninteresting_ref, &revs);
|
2012-05-04 20:14:48 +02:00
|
|
|
add_pending_sha1(&revs, "HEAD", new->object.sha1, UNINTERESTING);
|
2011-10-01 17:51:39 +02:00
|
|
|
|
2011-10-01 18:09:36 +02:00
|
|
|
refs = revs.pending;
|
|
|
|
revs.leak_pending = 1;
|
|
|
|
|
2011-02-19 01:04:47 +01:00
|
|
|
if (prepare_revision_walk(&revs))
|
2011-04-02 02:55:55 +02:00
|
|
|
die(_("internal error in revision walk"));
|
2012-05-04 20:14:48 +02:00
|
|
|
if (!(old->object.flags & UNINTERESTING))
|
|
|
|
suggest_reattach(old, &revs);
|
2011-02-19 01:04:47 +01:00
|
|
|
else
|
2012-05-04 20:14:48 +02:00
|
|
|
describe_detached_head(_("Previous HEAD position was"), old);
|
2011-03-20 10:09:18 +01:00
|
|
|
|
2011-10-01 18:16:08 +02:00
|
|
|
clear_commit_marks_for_object_array(&refs, ALL_REV_FLAGS);
|
2011-10-01 18:09:36 +02:00
|
|
|
free(refs.objects);
|
2011-02-19 01:04:47 +01:00
|
|
|
}
|
|
|
|
|
2012-08-29 15:55:23 +02:00
|
|
|
static int switch_branches(const struct checkout_opts *opts,
|
|
|
|
struct branch_info *new)
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
{
|
|
|
|
int ret = 0;
|
|
|
|
struct branch_info old;
|
2011-12-13 15:17:48 +01:00
|
|
|
void *path_to_free;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
unsigned char rev[20];
|
2012-08-29 15:55:22 +02:00
|
|
|
int flag, writeout_error = 0;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
memset(&old, 0, sizeof(old));
|
2014-07-15 21:59:36 +02:00
|
|
|
old.path = path_to_free = resolve_refdup("HEAD", 0, rev, &flag);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
old.commit = lookup_commit_reference_gently(rev, 1);
|
2011-12-13 15:17:48 +01:00
|
|
|
if (!(flag & REF_ISSYMREF))
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
old.path = NULL;
|
|
|
|
|
2014-06-18 21:48:29 +02:00
|
|
|
if (old.path)
|
|
|
|
skip_prefix(old.path, "refs/heads/", &old.name);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
|
|
|
if (!new->name) {
|
|
|
|
new->name = "HEAD";
|
|
|
|
new->commit = old.commit;
|
|
|
|
if (!new->commit)
|
2011-02-23 00:41:39 +01:00
|
|
|
die(_("You are on a branch yet to be born"));
|
2013-10-24 10:53:46 +02:00
|
|
|
parse_commit_or_die(new->commit);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
2012-08-29 15:55:22 +02:00
|
|
|
ret = merge_working_tree(opts, &old, new, &writeout_error);
|
2011-12-13 15:17:48 +01:00
|
|
|
if (ret) {
|
|
|
|
free(path_to_free);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
return ret;
|
2011-12-13 15:17:48 +01:00
|
|
|
}
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2009-05-16 19:54:45 +02:00
|
|
|
if (!opts->quiet && !old.path && old.commit && new->commit != old.commit)
|
2012-05-04 20:14:48 +02:00
|
|
|
orphaned_commit_warning(old.commit, new->commit);
|
2009-05-16 19:54:45 +02:00
|
|
|
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
update_refs_for_switch(opts, &old, new);
|
|
|
|
|
2008-05-29 00:26:59 +02:00
|
|
|
ret = post_checkout_hook(old.commit, new->commit, 1);
|
2011-12-13 15:17:48 +01:00
|
|
|
free(path_to_free);
|
2012-08-29 15:55:22 +02:00
|
|
|
return ret || writeout_error;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
2008-08-30 16:52:24 +02:00
|
|
|
static int git_checkout_config(const char *var, const char *value, void *cb)
|
|
|
|
{
|
2010-08-25 12:34:57 +02:00
|
|
|
if (!strcmp(var, "diff.ignoresubmodules")) {
|
|
|
|
struct checkout_opts *opts = cb;
|
|
|
|
handle_ignore_submodules_arg(&opts->diff_options, value);
|
|
|
|
return 0;
|
|
|
|
}
|
2010-08-28 16:50:07 +02:00
|
|
|
|
2013-11-30 21:55:40 +01:00
|
|
|
if (starts_with(var, "submodule."))
|
2010-08-28 16:50:07 +02:00
|
|
|
return parse_submodule_config_option(var, value);
|
|
|
|
|
2010-08-25 12:34:57 +02:00
|
|
|
return git_xmerge_config(var, value, NULL);
|
2008-08-30 16:52:24 +02:00
|
|
|
}
|
|
|
|
|
2009-10-18 09:13:47 +02:00
|
|
|
struct tracking_name_data {
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
/* const */ char *src_ref;
|
|
|
|
char *dst_ref;
|
|
|
|
unsigned char *dst_sha1;
|
2009-10-18 09:13:47 +02:00
|
|
|
int unique;
|
|
|
|
};
|
|
|
|
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
static int check_tracking_name(struct remote *remote, void *cb_data)
|
2009-10-18 09:13:47 +02:00
|
|
|
{
|
|
|
|
struct tracking_name_data *cb = cb_data;
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
struct refspec query;
|
|
|
|
memset(&query, 0, sizeof(struct refspec));
|
|
|
|
query.src = cb->src_ref;
|
|
|
|
if (remote_find_tracking(remote, &query) ||
|
2013-06-18 03:40:49 +02:00
|
|
|
get_sha1(query.dst, cb->dst_sha1)) {
|
|
|
|
free(query.dst);
|
2009-10-18 09:13:47 +02:00
|
|
|
return 0;
|
2013-06-18 03:40:49 +02:00
|
|
|
}
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
if (cb->dst_ref) {
|
2013-06-18 03:40:49 +02:00
|
|
|
free(query.dst);
|
2009-10-18 09:13:47 +02:00
|
|
|
cb->unique = 0;
|
|
|
|
return 0;
|
|
|
|
}
|
2013-06-18 03:40:49 +02:00
|
|
|
cb->dst_ref = query.dst;
|
2009-10-18 09:13:47 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
static const char *unique_tracking_name(const char *name, unsigned char *sha1)
|
2009-10-18 09:13:47 +02:00
|
|
|
{
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
struct tracking_name_data cb_data = { NULL, NULL, NULL, 1 };
|
|
|
|
char src_ref[PATH_MAX];
|
|
|
|
snprintf(src_ref, PATH_MAX, "refs/heads/%s", name);
|
|
|
|
cb_data.src_ref = src_ref;
|
|
|
|
cb_data.dst_sha1 = sha1;
|
|
|
|
for_each_remote(check_tracking_name, &cb_data);
|
2009-10-18 09:13:47 +02:00
|
|
|
if (cb_data.unique)
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
return cb_data.dst_ref;
|
|
|
|
free(cb_data.dst_ref);
|
2009-10-18 09:13:47 +02:00
|
|
|
return NULL;
|
|
|
|
}
|
2009-08-15 13:48:30 +02:00
|
|
|
|
2011-02-08 11:29:09 +01:00
|
|
|
static int parse_branchname_arg(int argc, const char **argv,
|
|
|
|
int dwim_new_local_branch_ok,
|
|
|
|
struct branch_info *new,
|
|
|
|
struct tree **source_tree,
|
|
|
|
unsigned char rev[20],
|
|
|
|
const char **new_branch)
|
|
|
|
{
|
|
|
|
int argcount = 0;
|
|
|
|
unsigned char branch_rev[20];
|
|
|
|
const char *arg;
|
2013-10-18 11:25:58 +02:00
|
|
|
int dash_dash_pos;
|
|
|
|
int has_dash_dash = 0;
|
|
|
|
int i;
|
2011-02-08 11:29:09 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
* 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.
|
|
|
|
*
|
2013-10-18 11:25:57 +02:00
|
|
|
* case 3: git checkout <something> [--]
|
2011-02-08 11:29:09 +01:00
|
|
|
*
|
2013-10-18 11:25:57 +02:00
|
|
|
* (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.
|
2011-02-08 11:29:09 +01:00
|
|
|
*
|
2013-10-18 11:25:57 +02:00
|
|
|
* (b) If <something> is _not_ a commit, either "--" is present
|
2014-04-01 00:11:46 +02:00
|
|
|
* or <something> is not a path, no -t or -b was given, and
|
2013-10-18 11:25:57 +02:00
|
|
|
* and there is a tracking branch whose name is <something>
|
|
|
|
* in one and only one remote, then this is a short-hand to
|
|
|
|
* fork local <something> from that remote-tracking branch.
|
2011-02-08 11:29:09 +01:00
|
|
|
*
|
2013-10-18 11:25:57 +02:00
|
|
|
* (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.
|
2011-02-08 11:29:09 +01:00
|
|
|
* - 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;
|
|
|
|
|
|
|
|
arg = argv[0];
|
2013-10-18 11:25:58 +02:00
|
|
|
dash_dash_pos = -1;
|
|
|
|
for (i = 0; i < argc; i++) {
|
|
|
|
if (!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);
|
2011-02-08 11:29:09 +01:00
|
|
|
|
|
|
|
if (!strcmp(arg, "-"))
|
|
|
|
arg = "@{-1}";
|
|
|
|
|
|
|
|
if (get_sha1_mb(arg, rev)) {
|
2013-10-18 11:25:57 +02:00
|
|
|
/*
|
|
|
|
* 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;
|
|
|
|
|
|
|
|
if (check_filename(NULL, arg) && !has_dash_dash)
|
|
|
|
recover_with_dwim = 0;
|
|
|
|
/*
|
|
|
|
* Accept "git checkout foo" and "git checkout foo --"
|
|
|
|
* as candidates for dwim.
|
|
|
|
*/
|
|
|
|
if (!(argc == 1 && !has_dash_dash) &&
|
|
|
|
!(argc == 2 && has_dash_dash))
|
|
|
|
recover_with_dwim = 0;
|
|
|
|
|
|
|
|
if (recover_with_dwim) {
|
checkout: Use remote refspecs when DWIMming tracking branches
The DWIM mode of checkout allows you to run "git checkout foo" when there
is no existing local ref or path called "foo", and there is exactly _one_
remote with a remote-tracking branch called "foo". Git will automatically
create a new local branch called "foo" using the remote-tracking "foo" as
its starting point and configured upstream.
For example, consider the following unconventional (but perfectly valid)
remote setup:
[remote "origin"]
fetch = refs/heads/*:refs/remotes/origin/*
[remote "frotz"]
fetch = refs/heads/*:refs/remotes/frotz/nitfol/*
Case 1: Assume both "origin" and "frotz" have remote-tracking branches called
"foo", at "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo"
respectively. In this case "git checkout foo" should fail, because there is
more than one remote with a "foo" branch.
Case 2: Assume only "frotz" have a remote-tracking branch called "foo". In
this case "git checkout foo" should succeed, and create a local branch "foo"
from "refs/remotes/frotz/nitfol/foo", using remote branch "foo" from "frotz"
as its upstream.
The current code hardcodes the assumption that all remote-tracking branches
must match the "refs/remotes/$remote/*" pattern (which is true for remotes
with "conventional" refspecs, but not true for the "frotz" remote above).
When running "git checkout foo", the current code looks for exactly one ref
matching "refs/remotes/*/foo", hence in the above example, it fails to find
"refs/remotes/frotz/nitfol/foo", which causes it to fail both case #1 and #2.
The better way to handle the above example is to actually study the fetch
refspecs to deduce the candidate remote-tracking branches for "foo"; i.e.
assume "foo" is a remote branch being fetched, and then map "refs/heads/foo"
through the refspecs in order to get the corresponding remote-tracking
branches "refs/remotes/origin/foo" and "refs/remotes/frotz/nitfol/foo".
Finally we check which of these happens to exist in the local repo, and
if there is exactly one, we have an unambiguous match for "git checkout foo",
and may proceed.
This fixes most of the failing tests introduced in the previous patch.
Signed-off-by: Johan Herland <johan@herland.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2013-04-21 23:52:01 +02:00
|
|
|
const char *remote = unique_tracking_name(arg, rev);
|
2013-10-18 11:25:57 +02:00
|
|
|
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);
|
2011-02-08 11:29:09 +01:00
|
|
|
return argcount;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* we can't end up being in (2) anymore, eat the argument */
|
|
|
|
argcount++;
|
|
|
|
argv++;
|
|
|
|
argc--;
|
|
|
|
|
|
|
|
new->name = arg;
|
|
|
|
setup_branch_path(new);
|
|
|
|
|
2011-09-15 23:10:25 +02:00
|
|
|
if (!check_refname_format(new->path, 0) &&
|
2011-11-13 11:22:14 +01:00
|
|
|
!read_ref(new->path, branch_rev))
|
2011-02-08 11:29:09 +01:00
|
|
|
hashcpy(rev, branch_rev);
|
|
|
|
else
|
|
|
|
new->path = NULL; /* not an existing branch */
|
|
|
|
|
|
|
|
new->commit = lookup_commit_reference_gently(rev, 1);
|
|
|
|
if (!new->commit) {
|
|
|
|
/* not a commit */
|
|
|
|
*source_tree = parse_tree_indirect(rev);
|
|
|
|
} else {
|
2013-10-24 10:53:46 +02:00
|
|
|
parse_commit_or_die(new->commit);
|
2011-02-08 11:29:09 +01:00
|
|
|
*source_tree = new->commit->tree;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!*source_tree) /* case (1): want a tree */
|
2011-04-02 02:55:55 +02:00
|
|
|
die(_("reference is not a tree: %s"), arg);
|
2013-10-18 11:25:57 +02:00
|
|
|
if (!has_dash_dash) {/* case (3).(d) -> (1) */
|
2011-02-08 11:29:09 +01:00
|
|
|
/*
|
|
|
|
* 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(NULL, arg);
|
|
|
|
} else {
|
|
|
|
argcount++;
|
|
|
|
argv++;
|
|
|
|
argc--;
|
|
|
|
}
|
|
|
|
|
|
|
|
return argcount;
|
|
|
|
}
|
|
|
|
|
2012-08-29 15:55:22 +02:00
|
|
|
static int switch_unborn_to_new_branch(const struct checkout_opts *opts)
|
2012-01-30 21:10:08 +01:00
|
|
|
{
|
|
|
|
int status;
|
|
|
|
struct strbuf branch_ref = STRBUF_INIT;
|
|
|
|
|
2012-06-26 17:06:42 +02:00
|
|
|
if (!opts->new_branch)
|
|
|
|
die(_("You are on a branch yet to be born"));
|
2012-01-30 21:10:08 +01:00
|
|
|
strbuf_addf(&branch_ref, "refs/heads/%s", opts->new_branch);
|
|
|
|
status = create_symref("HEAD", branch_ref.buf, "checkout -b");
|
|
|
|
strbuf_release(&branch_ref);
|
2012-11-15 18:12:33 +01:00
|
|
|
if (!opts->quiet)
|
|
|
|
fprintf(stderr, _("Switched to a new branch '%s'\n"),
|
|
|
|
opts->new_branch);
|
2012-01-30 21:10:08 +01:00
|
|
|
return status;
|
|
|
|
}
|
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
static int checkout_branch(struct checkout_opts *opts,
|
|
|
|
struct branch_info *new)
|
|
|
|
{
|
2013-07-14 10:35:41 +02:00
|
|
|
if (opts->pathspec.nr)
|
2012-08-30 14:45:50 +02:00
|
|
|
die(_("paths cannot be used with switching branches"));
|
|
|
|
|
|
|
|
if (opts->patch_mode)
|
|
|
|
die(_("'%s' cannot be used with switching branches"),
|
|
|
|
"--patch");
|
|
|
|
|
|
|
|
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->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");
|
|
|
|
} 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->name && !new->commit)
|
|
|
|
die(_("Cannot switch branch to a non-commit '%s'"),
|
|
|
|
new->name);
|
|
|
|
|
|
|
|
if (!new->commit && opts->new_branch) {
|
|
|
|
unsigned char rev[20];
|
|
|
|
int flag;
|
|
|
|
|
2014-07-15 21:59:36 +02:00
|
|
|
if (!read_ref_full("HEAD", 0, rev, &flag) &&
|
2012-08-30 14:45:50 +02:00
|
|
|
(flag & REF_ISSYMREF) && is_null_sha1(rev))
|
|
|
|
return switch_unborn_to_new_branch(opts);
|
|
|
|
}
|
|
|
|
return switch_branches(opts, new);
|
|
|
|
}
|
|
|
|
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
int cmd_checkout(int argc, const char **argv, const char *prefix)
|
|
|
|
{
|
|
|
|
struct checkout_opts opts;
|
|
|
|
struct branch_info new;
|
2008-09-01 04:32:40 +02:00
|
|
|
char *conflict_style = NULL;
|
2009-10-18 09:49:23 +02:00
|
|
|
int dwim_new_local_branch = 1;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
struct option options[] = {
|
2012-08-20 14:31:59 +02:00
|
|
|
OPT__QUIET(&opts.quiet, N_("suppress progress reporting")),
|
|
|
|
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")),
|
2013-08-03 13:51:19 +02:00
|
|
|
OPT_BOOL('l', NULL, &opts.new_branch_log, N_("create reflog for new branch")),
|
|
|
|
OPT_BOOL(0, "detach", &opts.force_detach, N_("detach the HEAD at named commit")),
|
2012-08-20 14:31:59 +02:00
|
|
|
OPT_SET_INT('t', "track", &opts.track, N_("set upstream info for new branch"),
|
2008-02-19 17:24:37 +01:00
|
|
|
BRANCH_TRACK_EXPLICIT),
|
2014-03-23 23:58:12 +01:00
|
|
|
OPT_STRING(0, "orphan", &opts.new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
|
2012-08-20 14:31:59 +02:00
|
|
|
OPT_SET_INT('2', "ours", &opts.writeout_stage, N_("checkout our version for unmerged files"),
|
2008-08-30 16:48:18 +02:00
|
|
|
2),
|
2012-08-20 14:31:59 +02:00
|
|
|
OPT_SET_INT('3', "theirs", &opts.writeout_stage, N_("checkout their version for unmerged files"),
|
2008-08-30 16:48:18 +02:00
|
|
|
3),
|
2012-08-20 14:31:59 +02:00
|
|
|
OPT__FORCE(&opts.force, N_("force checkout (throw away local modifications)")),
|
2013-08-03 13:51:19 +02:00
|
|
|
OPT_BOOL('m', "merge", &opts.merge, N_("perform a 3-way merge with the new branch")),
|
|
|
|
OPT_BOOL(0, "overwrite-ignore", &opts.overwrite_ignore, N_("update ignored files (default)")),
|
2012-08-20 14:31:59 +02:00
|
|
|
OPT_STRING(0, "conflict", &conflict_style, N_("style"),
|
|
|
|
N_("conflict style (merge or diff3)")),
|
2013-08-03 13:51:19 +02:00
|
|
|
OPT_BOOL('p', "patch", &opts.patch_mode, N_("select hunks interactively")),
|
2013-04-13 01:12:08 +02:00
|
|
|
OPT_BOOL(0, "ignore-skip-worktree-bits", &opts.ignore_skipworktree,
|
|
|
|
N_("do not limit pathspecs to sparse entries only")),
|
2013-08-03 13:51:18 +02:00
|
|
|
OPT_HIDDEN_BOOL(0, "guess", &dwim_new_local_branch,
|
|
|
|
N_("second guess 'git checkout no-such-branch'")),
|
2008-02-18 11:20:20 +01:00
|
|
|
OPT_END(),
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
};
|
|
|
|
|
|
|
|
memset(&opts, 0, sizeof(opts));
|
|
|
|
memset(&new, 0, sizeof(new));
|
2011-11-27 11:15:33 +01:00
|
|
|
opts.overwrite_ignore = 1;
|
2012-08-29 15:55:23 +02:00
|
|
|
opts.prefix = prefix;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2010-08-28 16:50:07 +02:00
|
|
|
gitmodules_config();
|
2010-08-25 12:34:57 +02:00
|
|
|
git_config(git_checkout_config, &opts);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2008-08-21 19:23:20 +02:00
|
|
|
opts.track = BRANCH_TRACK_UNSPECIFIED;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2009-05-23 20:53:12 +02:00
|
|
|
argc = parse_options(argc, argv, prefix, options, checkout_usage,
|
2008-07-23 12:15:32 +02:00
|
|
|
PARSE_OPT_KEEP_DASHDASH);
|
2008-07-23 12:15:33 +02:00
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
if (conflict_style) {
|
|
|
|
opts.merge = 1; /* implied */
|
|
|
|
git_xmerge_config("merge.conflictstyle", conflict_style, NULL);
|
|
|
|
}
|
2010-06-23 21:29:00 +02:00
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
if ((!!opts.new_branch + !!opts.new_branch_force + !!opts.new_orphan_branch) > 1)
|
|
|
|
die(_("-b, -B and --orphan are mutually exclusive"));
|
2010-06-23 21:29:00 +02:00
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
/*
|
|
|
|
* 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/--orphan is being used.
|
|
|
|
*/
|
2010-06-23 21:29:00 +02:00
|
|
|
if (opts.new_branch_force)
|
|
|
|
opts.new_branch = opts.new_branch_force;
|
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
if (opts.new_orphan_branch)
|
|
|
|
opts.new_branch = opts.new_orphan_branch;
|
2011-02-08 11:32:49 +01:00
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
/* --track without -b/-B/--orphan should DWIM */
|
|
|
|
if (opts.track != BRANCH_TRACK_UNSPECIFIED && !opts.new_branch) {
|
2008-08-21 19:23:20 +02:00
|
|
|
const char *argv0 = argv[0];
|
|
|
|
if (!argc || !strcmp(argv0, "--"))
|
2011-02-23 00:41:39 +01:00
|
|
|
die (_("--track needs a branch name"));
|
2014-10-04 20:54:50 +02:00
|
|
|
skip_prefix(argv0, "refs/", &argv0);
|
|
|
|
skip_prefix(argv0, "remotes/", &argv0);
|
2008-08-21 19:23:20 +02:00
|
|
|
argv0 = strchr(argv0, '/');
|
|
|
|
if (!argv0 || !argv0[1])
|
2011-02-23 00:41:39 +01:00
|
|
|
die (_("Missing branch name; try -b"));
|
2008-08-21 19:23:20 +02:00
|
|
|
opts.new_branch = argv0 + 1;
|
2008-08-09 16:00:12 +02:00
|
|
|
}
|
|
|
|
|
2008-07-23 12:15:33 +02:00
|
|
|
/*
|
2011-02-08 11:29:09 +01:00
|
|
|
* Extract branch name from command line arguments, so
|
|
|
|
* all that is left is pathspecs.
|
2008-07-23 12:15:33 +02:00
|
|
|
*
|
2011-02-08 11:29:09 +01:00
|
|
|
* Handle
|
2009-10-18 09:13:47 +02:00
|
|
|
*
|
2011-02-08 11:29:09 +01:00
|
|
|
* 1) git checkout <tree> -- [<paths>]
|
|
|
|
* 2) git checkout -- [<paths>]
|
|
|
|
* 3) git checkout <something> [<paths>]
|
2008-07-23 12:15:33 +02:00
|
|
|
*
|
2011-02-08 11:29:09 +01:00
|
|
|
* including "last branch" syntax and DWIM-ery for names of
|
|
|
|
* remote branches, erroring out for invalid or ambiguous cases.
|
2008-07-23 12:15:33 +02:00
|
|
|
*/
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
if (argc) {
|
2012-08-29 15:55:23 +02:00
|
|
|
unsigned char rev[20];
|
2011-02-08 11:29:09 +01:00
|
|
|
int dwim_ok =
|
2012-08-29 15:55:23 +02:00
|
|
|
!opts.patch_mode &&
|
2011-02-08 11:29:09 +01:00
|
|
|
dwim_new_local_branch &&
|
|
|
|
opts.track == BRANCH_TRACK_UNSPECIFIED &&
|
|
|
|
!opts.new_branch;
|
2011-02-08 11:34:34 +01:00
|
|
|
int n = parse_branchname_arg(argc, argv, dwim_ok,
|
2012-08-29 15:55:23 +02:00
|
|
|
&new, &opts.source_tree,
|
|
|
|
rev, &opts.new_branch);
|
2011-02-08 11:29:09 +01:00
|
|
|
argv += n;
|
|
|
|
argc -= n;
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
if (argc) {
|
2013-07-14 10:35:50 +02:00
|
|
|
parse_pathspec(&opts.pathspec, 0,
|
|
|
|
opts.patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
|
|
|
|
prefix, argv);
|
2008-02-28 17:30:47 +01:00
|
|
|
|
2013-07-14 10:35:41 +02:00
|
|
|
if (!opts.pathspec.nr)
|
2011-02-23 00:41:39 +01:00
|
|
|
die(_("invalid path specification"));
|
2008-02-28 17:30:47 +01:00
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
/*
|
|
|
|
* Try to give more helpful suggestion.
|
|
|
|
* new_branch && argc > 1 will be caught later.
|
|
|
|
*/
|
|
|
|
if (opts.new_branch && argc == 1)
|
|
|
|
die(_("Cannot update paths and switch to branch '%s' at the same time.\n"
|
|
|
|
"Did you intend to checkout '%s' which can not be resolved as commit?"),
|
|
|
|
opts.new_branch, argv[0]);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
|
2011-02-08 11:32:49 +01:00
|
|
|
if (opts.force_detach)
|
2012-08-30 14:45:50 +02:00
|
|
|
die(_("git checkout: --detach does not take a path argument '%s'"),
|
|
|
|
argv[0]);
|
2011-02-08 11:32:49 +01:00
|
|
|
|
2008-08-30 16:52:24 +02:00
|
|
|
if (1 < !!opts.writeout_stage + !!opts.force + !!opts.merge)
|
2012-08-30 14:45:50 +02:00
|
|
|
die(_("git checkout: --ours/--theirs, --force and --merge are incompatible when\n"
|
|
|
|
"checking out of the index."));
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|
|
|
|
|
2008-09-21 20:36:06 +02:00
|
|
|
if (opts.new_branch) {
|
2008-10-09 21:12:12 +02:00
|
|
|
struct strbuf buf = STRBUF_INIT;
|
2011-08-20 23:49:48 +02:00
|
|
|
|
2012-08-30 14:45:50 +02:00
|
|
|
opts.branch_exists =
|
|
|
|
validate_new_branchname(opts.new_branch, &buf,
|
|
|
|
!!opts.new_branch_force,
|
|
|
|
!!opts.new_branch_force);
|
2011-08-20 23:49:48 +02:00
|
|
|
|
2008-09-21 20:36:06 +02:00
|
|
|
strbuf_release(&buf);
|
|
|
|
}
|
|
|
|
|
2013-07-14 10:35:41 +02:00
|
|
|
if (opts.patch_mode || opts.pathspec.nr)
|
2012-08-30 14:45:50 +02:00
|
|
|
return checkout_paths(&opts, new.name);
|
|
|
|
else
|
|
|
|
return checkout_branch(&opts, &new);
|
Build in checkout
The only differences in behavior should be:
- git checkout -m with non-trivial merging won't print out
merge-recursive messages (see the change in t7201-co.sh)
- git checkout -- paths... will give a sensible error message if
HEAD is invalid as a commit.
- some intermediate states which were written to disk in the shell
version (in particular, index states) are only kept in memory in
this version, and therefore these can no longer be revealed by
later write operations becoming impossible.
- when we change branches, we discard MERGE_MSG, SQUASH_MSG, and
rr-cache/MERGE_RR, like reset always has.
I'm not 100% sure I got the merge recursive setup exactly right; the
base for a non-trivial merge in the shell code doesn't seem
theoretically justified to me, but I tried to match it anyway, and the
tests all pass this way.
Other than these items, the results should be identical to the shell
version, so far as I can tell.
[jc: squashed lock-file fix from Dscho in]
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2008-02-07 17:40:23 +01:00
|
|
|
}
|