"git pull" learned to pay attention to pull.ff configuration
variable.
* da/pull-ff-configuration:
pull: add --ff-only to the help text
pull: add pull.ff configuration
Improvements to our hash table to get it to meet the needs of the
msysgit fscache project, with some nice performance improvements.
* kb/fast-hashmap:
name-hash: retire unused index_name_exists()
hashmap.h: use 'unsigned int' for hash-codes everywhere
test-hashmap.c: drop unnecessary #includes
.gitignore: test-hashmap is a generated file
read-cache.c: fix memory leaks caused by removed cache entries
builtin/update-index.c: cleanup update_one
fix 'git update-index --verbose --again' output
remove old hash.[ch] implementation
name-hash.c: remove cache entries instead of marking them CE_UNHASHED
name-hash.c: use new hash map implementation for cache entries
name-hash.c: remove unreferenced directory entries
name-hash.c: use new hash map implementation for directories
diffcore-rename.c: use new hash map implementation
diffcore-rename.c: simplify finding exact renames
diffcore-rename.c: move code around to prepare for the next patch
buitin/describe.c: use new hash map implementation
add a hashtable implementation that supports O(1) removal
submodule: don't access the .gitmodules cache entry after removing it
Introduce commit.gpgsign configuration variable to force every
commit to be GPG signed. The variable cannot be overriden from the
command line of some of the commands that create commits except for
"git commit" and "git commit-tree", but I am not convinced that it
is a good idea to sprinkle support for --no-gpg-sign everywhere,
which in turn means that this configuration variable may not be
such a good idea.
* nv/commit-gpgsign-config:
test the commit.gpgsign config option
commit-tree: add and document --no-gpg-sign
commit-tree: add the commit.gpgsign option to sign all commits
We sometimes write tempfiles of the form "shallow_XXXXXX"
during fetch/push operations with shallow repositories.
Under normal circumstances, we clean up the result when we
are done. However, we do no take steps to clean up after
ourselves when we exit due to die() or signal death.
This patch teaches the tempfile creation code to register
handlers to clean up after ourselves. To handle this, we
change the ownership semantics of the filename returned by
setup_temporary_shallow. It now keeps a copy of the filename
itself, and returns only a const pointer to it.
We can also do away with explicit tempfile removal in the
callers. They all exit not long after finishing with the
file, so they can rely on the auto-cleanup, simplifying the
code.
Note that we keep things simple and maintain only a single
filename to be cleaned. This is sufficient for the current
caller, but we future-proof it with a die("BUG").
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we are about to write the shallow file, we check that
it has not changed since we last read it. Instead of
hand-rolling this, we can use stat_validity. This is built
around the index stat-check, so it is more robust than just
checking the mtime, as we do now (it uses the same check as
we do for index files).
The new code also handles the case of a shallow file
appearing unexpectedly. With the current code, two
simultaneous processes making us shallow (e.g., two "git
fetch --depth=1" running at the same time in a non-shallow
repository) can race to overwrite each other.
As a bonus, we also remove a race in determining the stat
information of what we read (we stat and then open, leaving
a race window; instead we should open and then fstat the
descriptor).
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor binary search in "commit_graft_pos" function: use
generic "sha1_pos" function.
Signed-off-by: Dmitry S. Dolzhenko <dmitrys.dolzhenko@yandex.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When asking to fetch/pull a branch whose name is B or a tag whose
name is T, we used to show the command to run as:
git pull $URL B
git pull $URL tags/T
even when B and T were spelled in a more qualified way in order to
disambiguate, e.g. heads/B or refs/tags/T, but the recent update
lost this feature. Resurrect it.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This illustrates behaviour changes that result from the recent
change by Linus. Most show good changes, but there may be some
usability regressions:
- The command continues to fail when the user forgot to push out
before running the command, but the wording of the message has
been slightly changed.
- The command no longer guesses when asked to request the commit at
the HEAD be pulled after pushing it to a branch 'for-upstream',
even when that branch points at the correct commit. The user
must ask the command with the new "master:for-upstream" syntax.
The new behaviour needs to be documented in any case, but we need to
agree what the new behaviour should be before doing so first.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous two steps were meant to stop updating the explicit
refname the user gave to the command to a different ref that points
at it. Most notably, we no longer substitute a branch name the user
used with a name of the tag that points at the commit at the tip of
the branch (it still can be done with "local-branch:remote-tag").
However, they also lost the code that included the message in a
tag when the user _did_ ask the tag to be pulled. Resurrect it.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This allows a user to say that a local branch has a different name on
the remote server, using the same syntax that "git push" uses to create
that situation.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The current 'request-pull' will try to find matching commit on the given
remote, and rewrite the "please pull" line to match that remote ref.
That may be very helpful if your local tree doesn't match the layout of
the remote branches, but for the common case it's been a recurring
disaster, when "request-pull" is done against a delayed remote update, and
it rewrites the target branch randomly to some other branch name that
happens to have the same expected SHA1 (or more commonly, leaves it
blank).
To avoid that recurring problem, this changes "git request-pull" so that
it matches the ref name to be pulled against the *local* repository, and
then warns if the remote repository does not have that exact same branch
or tag name and content.
This means that git request-pull will never rewrite the ref-name you gave
it. If the local branch name is "xyzzy", that is the only branch name
that request-pull will ask the other side to fetch.
If the remote has that branch under a different name, that's your problem
and git request-pull will not try to fix it up (but git request-pull will
warn about the fact that no exact matching branch is found, and you can
edit the end result to then have the remote name you want if it doesn't
match your local one).
The new "find local ref" code will also complain loudly if you give an
ambiguous refname (eg you have both a tag and a branch with that same
name, and you don't specify "heads/name" or "tags/name").
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since 1a72cfd (commit -v: strip diffs and submodule shortlogs from the
commit message - 2013-12-05) we have a less fragile way to cut out
"git status" at the end of a commit message but it's only enabled for
stripping submodule shortlogs.
Add new cleanup option that reuses the same mechanism for the entire
"git status" without accidentally removing lines starting with '#'.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
db5360f3f4 (name-hash: refactor polymorphic index_name_exists();
2013-09-17) split index_name_exists() into index_file_exists() and
index_dir_exists() but retained index_name_exists() as a thin wrapper
to avoid disturbing possible in-flight topics. Since this change
landed in 'master' some time ago and there are no in-flight topics
referencing index_name_exists(), retire it.
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The tests are checking that :
- when commit.gpgsign is true, "git commit" creates signed commits
- when commit.gpgsign is false, "git commit" creates unsigned commits
- when commit.gpgsign is true, "git commit --no-gpg-sign" creates
unsigned commits
- when commit.gpgsign is true, "git rebase -f" creates signed commits
Signed-off-by: Nicolas Vigier <boklm@mars-attacks.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Document how to override commit.gpgsign configuration that is set to
true per "git commit" invocation (parse-options machinery lets us
say "--no-gpg-sign" to do so).
"git commit-tree" does not use parse-options, so manually add the
corresponding option for now.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If you want to GPG sign all your commits, you have to add the -S option
all the time. The commit.gpgsign config option allows to sign all
commits automatically.
Signed-off-by: Nicolas Vigier <boklm@mars-attacks.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When QUICK is set (i.e. with --quiet) we try to do as little work as
possible, stopping after seeing the first change. stat-dirty is
considered a "change" but it may turn out not, if no actual content is
changed. The actual content test is performed too late in the process
and the shortcut may be taken prematurely, leading to incorrect return
code.
Assume we do "git diff --quiet". If we have a stat-dirty file "a" and
a really dirty file "b". We break the loop in run_diff_files() and
stop after "a" because we have got a "change". Later in
diffcore_skip_stat_unmatch() we find out "a" is actually not
changed. But there's nothing else in the diff queue, we incorrectly
declare "no change", ignoring the fact that "b" is changed.
This also happens to "git diff --quiet HEAD" when it hits
diff_can_quit_early() in oneway_diff().
This patch does the content test earlier in order to keep going if "a"
is unchanged. The test result is cached so that when
diffcore_skip_stat_unmatch() is done in the end, we spend no cycles on
re-testing "a".
Reported-by: IWAMOTO Toshihiro <iwamoto@valinux.co.jp>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
where "correct paths" stands for paths that are different to all
parents.
Up until now, we were testing combined diff only on one file, or on
several files which were all different (t4038-diff-combined.sh).
As recent thinko in "simplify intersect_paths() further" showed, and
also, since we are going to rework code for finding paths different to
all parents, lets write at least basic tests.
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Linus once said:
I actually wish more people understood the really core low-level
kind of coding. Not big, complex stuff like the lockless name
lookup, but simply good use of pointers-to-pointers etc. For
example, I've seen too many people who delete a singly-linked
list entry by keeping track of the "prev" entry, and then to
delete the entry, doing something like
if (prev)
prev->next = entry->next;
else
list_head = entry->next;
and whenever I see code like that, I just go "This person
doesn't understand pointers". And it's sadly quite common.
People who understand pointers just use a "pointer to the entry
pointer", and initialize that with the address of the
list_head. And then as they traverse the list, they can remove
the entry without using any conditionals, by just doing a "*pp =
entry->next".
Applying that simplification lets us lose 7 lines from this function
even while adding 2 lines of comment.
I was tempted to squash this into the original commit, but because
the benchmarking described in the commit log is without this
simplification, I decided to keep it a separate follow-up patch.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The field was used in order to speed-up name comparison and also to
mark removed paths by setting it to 0.
Because the updated code does significantly less strcmp and also
just removes paths from the list and free right after we know a path
will not be needed, it is not needed anymore.
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When generating combined diff, for each commit, we intersect diff
paths from diff(parent_0,commit) to diff(parent_i,commit) comparing
all paths pairs, i.e. doing it the quadratic way. That is correct,
but could be optimized.
Paths come from trees in sorted (= tree) order, and so does diff_tree()
emits resulting paths in that order too. Now if we look at diffcore
transformations, all of them, except diffcore_order, preserve resulting
path ordering:
- skip_stat_unmatch, grep, pickaxe, filter
-- just skip elements -> order stays preserved
- break -- just breaks diff for a path, adding path
dup after the path -> order stays preserved
- detect rename/copy -- resulting paths are emitted sorted
(verified empirically)
So only diffcore_order changes diff paths ordering.
But diffcore_order meaning affects only presentation - i.e. only how to
show the diff, so we could do all the internal computations without
paths reordering, and order only resultant paths set. This is faster,
since, if we know two paths sets are all ordered, their intersection
could be done in linear time.
This patch does just that.
Timings for `git log --raw --no-abbrev --no-renames` without `-c` ("git log")
and with `-c` ("git log -c") before and after the patch are as follows:
linux.git v3.10..v3.11
log log -c
before 1.9s 20.4s
after 1.9s 16.6s
navy.git (private repo)
log log -c
before 0.83s 15.6s
after 0.83s 2.1s
P.S.
I think linux.git case is sped up not so much as the second one, since
in navy.git, there are more exotic (subtree, etc) merges.
P.P.S.
My tracing showed that the rest of the time (16.6s vs 1.9s) is usually
spent in computing huge diffs from commit to second parent. Will try to
deal with it, if I'll have time.
P.P.P.S.
For combine_diff_path, ->len is not needed anymore - will remove it in
the next noisy cleanup path, to maintain good signal/noise ratio here.
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In the next patch combine-diff will have special code-path for taking
orderfile into account. Prepare for making changes by introducing
coverage tests for that case.
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
diffcore_order() interface only accepts a queue of `struct
diff_filepair`.
In the next patches, we'll want to order `struct combine_diff_path`
by path, so let's first rework diffcore-order to also provide
generic low-level interface for ordering arbitrary objects, provided
they have path accessors.
The new interface is:
- `struct obj_order` for describing objects to ordering routine, and
- order_objects() for actually doing the ordering work.
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This continues 4651ece8 (Switch over tree descriptors to contain a
pre-parsed entry) and moves the only rest computational part
mode = canon_mode(mode)
from tree_entry_extract() to tree entry decode phase - to
decode_tree_entry().
The reason to do it, is that canon_mode() is at least 2 conditional
jumps for regular files, and that could be noticeable should canon_mode()
be invoked several times.
That does not matter for current Git codebase, where typical tree
traversal is
while (t->size) {
sha1 = tree_entry_extract(t, &path, &mode);
...
update_tree_entry(t);
}
i.e. we do t -> sha1,path.mode "extraction" only once per entry. In such
cases, it does not matter performance-wise, where that mode
canonicalization is done - either once in tree_entry_extract(), or once
in decode_tree_entry() called by update_tree_entry() - it is
approximately the same.
But for future code, which could need to work with several tree_desc's
in parallel, it could be handy to operate on tree_desc descriptors, and
do "extracts" only when needed, or at all, access only relevant part of
it through structure fields directly.
And for such situations, having canon_mode() be done once in decode
phase is better - we won't need to pay the performance price of 2 extra
conditional jumps on every t->mode access.
So let's move mode canonicalization to decode_tree_entry(). That was the
final bit. Now after tree entry is decoded, it is fully ready and could
be accessed either directly via field, or through tree_entry_extract()
which this time got really "totally trivial".
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
cmd_clean() has the exact same code of index_name_is_other(). Reduce
code duplication.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This instance was left out when many match_pathspec() call sites that
take input from dir_entry were converted to dir_path_match() because
it passed a path with the trailing slash stripped out to match_pathspec()
while the others did not. Stripping for all call sites back then would
be a regression because match_pathspec() did not know how to match
pathspec foo/ against _directory_ foo (the stripped version of path
"foo/").
match_pathspec() knows how to do it now. And dir_path_match() strips
the trailing slash also. Use the new function, because the stripping
code is removed in the next patch.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch activates the DO_MATCH_DIRECTORY code in m_p_i(), which
makes "git diff HEAD submodule/" and "git diff HEAD submodule" produce
the same output. Previously only the version without trailing slash
returns the difference (if any).
That's the effect of new ce_path_match(). dir_path_match() is not
executed by the new tests. And it should not introduce regressions.
Previously if path "dir/" is passed in with pathspec "dir/", they
obviously match. With new dir_path_match(), the path becomes
_directory_ "dir" vs pathspec "dir/", which is not executed by the old
code path in m_p_i(). The new code path is executed and produces the
same result.
The other case is pathspec "dir" and path "dir/" is now turned to
"dir" (with DO_MATCH_DIRECTORY). Still the same result before or after
the patch.
So why change? Because of the next patch about clean.c.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently we do support matching pathspec "foo/" against directory
"foo". That is because match_pathspec() has no way to tell "foo" is a
directory and matching "foo/" against _file_ "foo" is wrong.
The callers can now tell match_pathspec if "foo" is a directory, we
could make an exception for this case. Code is not executed though
because no callers pass the flag yet.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A long time ago, for some reason I was not happy with
match_pathspec(). I created a better version, match_pathspec_depth()
that was suppose to replace match_pathspec()
eventually. match_pathspec() has finally been gone since 6 months
ago. Use the shorter name for match_pathspec_depth().
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This helps reduce the number of match_pathspec_depth() call sites and
show how m_p_d() is used. And it usage is:
- match against an index entry (ce_path_match or match_pathspec_depth
in ls-files)
- match against a dir_entry from read_directory (dir_path_match and
match_pathspec_depth in clean.c, which will be converted later)
- resolve-undo (rerere.c and ls-files.c)
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This helps reduce the number of match_pathspec_depth() call sites and
show how match_pathspec_depth() is used.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make it clear that there is no implicit floating going on; --remote
lets you explicitly integrate the upstream branch in your current
HEAD (just like running 'git pull' in the submodule). The only
distinction with the current 'git pull' is the config location and
setting used for the upstream branch, which is hopefully clear now.
Signed-off-by: W. Trevor King <wking@tremily.us>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous code only checked out branches in cmd_add. This commit
moves the branch-checkout logic into module_clone, where it can be
shared by cmd_add and cmd_update. I also update the initial checkout
command to use 'reset' to preserve branches setup during module_clone.
With this change, folks cloning submodules for the first time via:
$ git submodule update ...
will get a local branch instead of a detached HEAD, unless they are
using the default checkout-mode updates. This is a change from the
previous situation where cmd_update always used checkout-mode logic
(regardless of the requested update mode) for updates that triggered
an initial clone, which always resulted in a detached HEAD.
This commit does not change the logic for updates after the initial
clone, which will continue to create detached HEADs for checkout-mode
updates, and integrate remote work with the local HEAD (detached or
not) in other modes.
The motivation for the change is that developers doing local work
inside the submodule are likely to select a non-checkout-mode for
updates so their local work is integrated with upstream work.
Developers who are not doing local submodule work stick with
checkout-mode updates so any apparently local work is blown away
during updates. For example, if upstream rolls back the remote branch
or gitlinked commit to an earlier version, the checkout-mode developer
wants their old submodule checkout to be rolled back as well, instead
of getting a no-op merge/rebase with the rolled-back reference.
By using the update mode to distinguish submodule developers from
black-box submodule consumers, we can setup local branches for the
developers who will want local branches, and stick with detached HEADs
for the developers that don't care.
Testing
=======
In t7406, just-cloned checkouts now update to the gitlinked hash with
'reset', to preserve the local branch for situations where we're not
on a detached HEAD.
I also added explicit tests to t7406 for HEAD attachement after
cloning updates, showing that it depends on their update mode:
* Checkout-mode updates get detached HEADs
* Everyone else gets a local branch, matching the configured
submodule.<name>.branch and defaulting to master.
The 'initial-setup' tag makes it easy to reset the superproject to a
known state, as several earlier tests commit to submodules and commit
the changed gitlinks to the superproject, but don't push the new
submodule commits to the upstream subprojects. This makes it
impossible to checkout the current super master, because it references
submodule commits that don't exist in the upstream subprojects. For a
specific example, see the tests that currently generate the
'two_new_submodule_commits' commits.
Documentation
=============
I updated the docs to describe the 'submodule update' modes in detail.
The old documentation did not distinguish between cloning and
non-cloning updates and lacked clarity on which operations would lead
to detached HEADs, and which would not. The new documentation
addresses these issues while updating the docs to reflect the changes
introduced by this commit's explicit local branch creation in
module_clone.
I also add '--checkout' to the usage summary and group the update-mode
options into a single set.
Signed-off-by: W. Trevor King <wking@tremily.us>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This avoids the current awkwardness of having either '' or 'checkout'
for checkout-mode updates, which makes testing for checkout-mode
updates (or non-checkout-mode updates) easier.
Signed-off-by: W. Trevor King <wking@tremily.us>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Per Documentation/CodingGuidelines most C files in git start with
a #include of git-compat-util.h or another header file that includes
it, such as cache.h or builtin.h. This file doesn't need anything
beyond "git-compat-util.h", so use that.
Remove a #include of the system header <stdio.h> since it is already
included by "git-compat-util.h".
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Prevent the "test-hashmap" program from being accidentally tracked
with "git add" or cluttering "git status" output.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Making a single preparation run for counting the lines will avoid memory
fragmentation. Also, fix the allocated memory size which was wrong
when sizeof(int *) != sizeof(int), and would have been too small
for sizeof(int *) < sizeof(int), admittedly unlikely.
Signed-off-by: David Kastrup <dak@gnu.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If we are calling xrealloc on every single line, the least we can do
is get the right allocation size.
Signed-off-by: David Kastrup <dak@gnu.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since the origin pointers are "interned" and reference-counted, comparing
the pointers rather than the content is enough. The only uninterned
origins are cached values kept in commit->util, but same_suspect is not
called on them.
Signed-off-by: David Kastrup <dak@gnu.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach add_cacheinfo to tell make_cache_entry to skip refreshing stat
information when a file is missing from the work tree. We do not want
the index to be stat-dirty after the merge but also do not want to fail
when a file happens to be missing.
This fixes the 'merge-recursive w/ empty work tree - ours has rename'
case in t3030-merge-recursive.
Suggested-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Brad King <brad.king@kitware.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Convert the make_cache_entry boolean 'refresh' argument to a more
general 'refresh_options' argument. Pass the value through to the
underlying refresh_cache_ent call. Add option CE_MATCH_REFRESH to
enable stat refresh. Update call sites to use the new signature.
Signed-off-by: Brad King <brad.king@kitware.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move lstat ENOENT handling from refresh_index to refresh_cache_ent and
activate it with a new CE_MATCH_IGNORE_MISSING option. This will allow
other call paths into refresh_cache_ent to use the feature.
Signed-off-by: Brad King <brad.king@kitware.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Sometimes when working with a large repository it can be useful to try
out a merge and only check out conflicting files to disk (for example as
a speed optimization on a server). Until v1.7.7-rc1~28^2~20
(merge-recursive: When we detect we can skip an update, actually skip
it, 2011-08-11), it was possible to do so with the following idiom:
# Prepare a temporary index and empty work tree.
GIT_INDEX_FILE="$PWD/tmp-$$-index" &&
export GIT_INDEX_FILE &&
GIT_WORK_TREE="$PWD/tmp-$$-work" &&
export GIT_WORK_TREE &&
mkdir "$GIT_WORK_TREE" &&
# Convince the index that our side is on disk.
git read-tree -i -m $ours &&
git update-index --ignore-missing --refresh &&
# Merge their side into our side.
bases=$(git merge-base --all $ours $theirs) &&
git merge-recursive $bases -- $ours $theirs &&
tree=$(git write-tree)
Nowadays, that still works and the exit status is the same, but
merge-recursive produces a diagnostic if "our" side renamed a file:
error: addinfo_cache failed for path 'dst'
Add a test to document this regression.
Signed-off-by: Brad King <brad.king@kitware.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
