* jc/cache-unmerge:
rerere forget path: forget recorded resolution
rerere: refactor rerere logic to make it independent from I/O
rerere: remove silly 1024-byte line limit
resolve-undo: teach "update-index --unresolve" to use resolve-undo info
resolve-undo: "checkout -m path" uses resolve-undo information
resolve-undo: allow plumbing to clear the information
resolve-undo: basic tests
resolve-undo: record resolved conflicts in a new index extension section
builtin-merge.c: use standard active_cache macros
Conflicts:
builtin-ls-files.c
builtin-merge.c
builtin-rerere.c
When switching to a different commit, we first see the named rev exists
as a commit using lookup_commit_reference_gently(), and set new.path to
a string "refs/heads/" followed by the name the user gave us (but after
taking into special short-hands like @{-1} == "previous branch" and
"@{upstream}" == "the branch we merge with" into account). If the
resulting string names an existsing ref, then we are switching to that
branch (and will be building new commits on top of it); otherwise we are
detaching HEAD at that commit.
When the "master..." syntax is used as a short-hand for "master...HEAD",
we do want to detach HEAD at the merge base. However, on Windows, when
asked if ".git/refs/heads/master..." exists, the filesystem happily says
"it does" when ".git/refs/heads/master" exists.
Work this issue around by first calling check_ref_format(new.path) to see
if the string can possibly be a valid ref under "refs/heads/", before
asking resolve_ref().
We used to run another lookup_commit_reference(rev) even though we know it
succeeded and we have a good commit in new.commit already; this has been
with us from 782c2d6 (Build in checkout, 2008-02-07), the first version we
had "git checkout" implemented in C. Drop it.
Noticed by Alex Riesen.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jc/checkout-merge-base:
rebase -i: teach --onto A...B syntax
rebase: fix --onto A...B parsing and add tests
"rebase --onto A...B" replays history on the merge base between A and B
"checkout A...B" switches to the merge base between A and B
* maint-1.6.2:
base85: Make the code more obvious instead of explaining the non-obvious
base85: encode_85() does not use the decode table
base85 debug code: Fix length byte calculation
checkout -m: do not try to fall back to --merge from an unborn branch
Conflicts:
diff.c
* maint-1.6.1:
base85: Make the code more obvious instead of explaining the non-obvious
base85: encode_85() does not use the decode table
base85 debug code: Fix length byte calculation
checkout -m: do not try to fall back to --merge from an unborn branch
branch: die explicitly why when calling "git branch [-a|-r] branchname".
textconv: stop leaking file descriptors
commit: --cleanup is a message option
git count-objects: handle packs bigger than 4G
t7102: make the test fail if one of its check fails
Conflicts:
diff.c
* maint-1.6.0:
base85: Make the code more obvious instead of explaining the non-obvious
base85: encode_85() does not use the decode table
base85 debug code: Fix length byte calculation
checkout -m: do not try to fall back to --merge from an unborn branch
branch: die explicitly why when calling "git branch [-a|-r] branchname".
If switching from an unborn branch (= empty tree) to a valid commit failed
without -m, it would fail with -m option as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We should tell ll_merge() that the 3-way merge between stages #2 and #3 is
an outermost merge, not a virtual-ancestor creation.
Back when this code was originally written, users couldn't write custom
merge drivers easily, so the bug didn't matter, but these days it does.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Once you resolved conflicts by "git add path", you cannot recreate the
conflicted state with "git checkout -m path", because you lost information
from higher stages in the index when you resolved them.
Since we record the necessary information in the resolve-undo index
extension these days, we can reproduce the unmerged state in the index and
check it out.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When resolving a conflict using "git add" to create a stage #0 entry, or
"git rm" to remove entries at higher stages, remove_index_entry_at()
function is eventually called to remove unmerged (i.e. higher stage)
entries from the index. Introduce a "resolve_undo_info" structure and
keep track of the removed cache entries, and save it in a new index
extension section in the index_state.
Operations like "read-tree -m", "merge", "checkout [-m] <branch>" and
"reset" are signs that recorded information in the index is no longer
necessary. The data is removed from the index extension when operations
start; they may leave conflicted entries in the index, and later user
actions like "git add" will record their conflicted states afresh.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Porcelains may want to make sure their calls to "git checkout" will
reliably fail regardless of the presense of random remote tracking
branches by the new DWIMmery introduced.
Luckily all existing in-tree callers have extra checks to make sure they
feed local branch name when they want to switch, or they explicitly ask to
detach HEAD at the given commit, so there is no need to add this option
for them.
As this is strictly script-only option, do not even bother to document it,
and do bother to hide it from "git checkout -h".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
pretty_print_commit() has a bunch of rarely-used arguments, and
introducing more of them requires yet another update of all the call
sites. Refactor most of them into a struct to make future extensions
easier.
The ones that stay "plain" arguments were chosen on the grounds that
all callers put real arguments there, whereas some callers have 0/NULL
for all arguments that were factored into the struct.
We declare the struct 'const' to ensure none of the callers are bitten
by the changed (no longer call-by-value) semantics.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When flipping commits around on topic branches, I often end up doing
this sequence:
* Run "log --oneline next..jc/frotz" to find out the first commit
on 'jc/frotz' branch not yet merged to 'next';
* Run "checkout $that_commit^" to detach HEAD to the parent of it;
* Rebuild the series on top of that commit; and
* "show-branch jc/frotz HEAD" and "diff jc/frotz HEAD" to verify.
Introduce a new syntax to "git checkout" to name the commit to switch to,
to make the first two steps easier. When the branch to switch to is
specified as A...B (you can omit either A or B but not both, and HEAD
is used instead of the omitted side), the merge base between these two
commits are computed, and if there is one unique one, we detach the HEAD
at that commit.
With this, I can say "checkout next...jc/frotz".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When 'frotz' is not a valid object name and not a tracked filename,
we used to complain and failed this command. When there is only
one remote that has 'frotz' as one of its tracking branches, we can
DWIM it as a request to create a local branch 'frotz' forking from
the matching remote tracking branch.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* tr/reset-checkout-patch:
stash: simplify defaulting to "save" and reject unknown options
Make test case number unique
tests: disable interactive hunk selection tests if perl is not available
DWIM 'git stash save -p' for 'git stash -p'
Implement 'git stash save --patch'
Implement 'git checkout --patch'
Implement 'git reset --patch'
builtin-add: refactor the meat of interactive_add()
Add a small patch-mode testing library
git-apply--interactive: Refactor patch mode code
Make 'git stash -k' a short form for 'git stash save --keep-index'
git branch, checkout, clean, mv and tag all have an option -f to override
certain checks. This patch makes them accept the long option --force as
a synonym.
While we're at it, document that checkout support --quiet as synonym for
its short option -q.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
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>
This introduces a --patch mode for git-checkout. In the index usage
git checkout --patch -- [files...]
it lets the user discard edits from the <files> at the granularity of
hunks (by selecting hunks from 'git diff' and then reverse applying
them to the worktree).
We also accept a revision argument. In the case
git checkout --patch HEAD -- [files...]
we offer hunks from the difference between HEAD and the worktree, and
reverse applies them to both index and worktree, allowing you to
discard staged changes completely. In the non-HEAD usage
git checkout --patch <revision> -- [files...]
it offers hunks from the difference between the worktree and
<revision>. The chosen hunks are then applied to both index and
worktree.
The application to worktree and index is done "atomically" in the
sense that we first check if the patch applies to the index (it should
always apply to the worktree). If it does not, we give the user a
choice to either abort or apply to the worktree anyway.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
To give OPT_FILENAME the prefix, we pass the prefix to parse_options()
which passes the prefix to parse_options_start() which sets the prefix
member of parse_opts_ctx accordingly. If there isn't a prefix in the
calling context, passing NULL will suffice.
Signed-off-by: Stephen Boyd <bebarino@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously, checkout would tell the user this message before moving HEAD,
without regard to whether the upcoming move will result in success.
If the move failed, this causes confusion.
Show the message after the move, unless the move failed.
Signed-off-by: Daniel Cordero <theappleman@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The logic in 83ae209 (checkout branch: prime cache-tree fully,
2009-04-20) is bogus; checkout can switch branches with a dirty
index and in such a case the tree won't match HEAD.
Add t2014-switch to catch this breakage.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This makes git checkout know to use the threaded index preloading if it
is enabled in the config file. You need to have
[core]
preloadindex = true
in your config file to see it, and for that feature to make sense your
filesystem needs to be able to do concurrent 'lstat()' lookups, but when
that is the case (especially NFS over a high-latency network), this can
be a noticeable performance win.
But with a low-latency network and at least older Linux NFS clients, this
will clearly potentially cause a lot of lock contention. It may still
speed up the uncached case, but the threading and locking overhead will
result in the cached case likely slowing down.
That was almost certainly fixed by Linux commit fc0f684c2 ("NFS: Remove
BKL from NFS lookup code"), but that one got merged into 2.6.27-rc1, so
older kernel versions than 2.6.27 will not scale very well.
But regardless, it's the right thing to do. If your filesystem doesn't
scale, don't enable index preloading.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jc/maint-read-tree-multi:
checkout branch: prime cache-tree fully
read-tree -m A B: prime cache-tree from the switched-to tree
Move prime_cache_tree() to cache-tree.c
read-tree A B: do not corrupt cache-tree
When switching to another branch, the earlier code relied on incremental
invalidation of cache-tree entries to degrade it. While it is not wrong
per-se, we know that the resulting index must fully match the branch we
are switching to unless the -m (merge) option is used.
We should simply fully re-prime the cache-tree using the new tree object
in such a case. And for safety, invalidate the cache-tree as a whole in
other cases.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint:
Change double quotes to single quotes in message
Documentation: clarify .gitattributes search
git-checkout.txt: clarify that <branch> applies when no path is given.
git-checkout.txt: fix incorrect statement about HEAD and index
Conflicts:
Documentation/git-checkout.txt
Most of the time when we give branch name in the message, we quote it
inside a pair of single-quotes. git-checkout uses double-quotes; this
patch corrects the inconsistency.
Signed-off-by: Jari Aalto <jari.aalto@cante.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jc/name-branch:
Don't permit ref/branch names to end with ".lock"
check_ref_format(): tighten refname rules
strbuf_check_branch_ref(): a helper to check a refname for a branch
Fix branch -m @{-1} newname
check-ref-format --branch: give Porcelain a way to grok branch shorthand
strbuf_branchname(): a wrapper for branch name shorthands
Rename interpret/substitute nth_last_branch functions
Conflicts:
Documentation/git-check-ref-format.txt
While 'git checkout <submodule>' should not update the submodule's
working directory, it should update the index. This is in line with
how submodules are handled in the rest of Git.
While at it, test 'git reset [<commit>] <submodule>', too.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This allows a common calling sequence
strbuf_branchname(&ref, name);
strbuf_splice(&ref, 0, 0, "refs/heads/", 11);
if (check_ref_format(ref.buf))
die(...);
to be refactored into
if (strbuf_check_branch_ref(&ref, name))
die(...);
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function takes a user-supplied string that is supposed to be a branch
name, and puts it in a strbuf after expanding possible shorthand notation.
A handful of open coded sequence to do this in the existing code have been
changed to use this helper function.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
These allow you to say "git checkout @{-2}" to switch to the branch two
"branch switching" ago by pretending as if you typed the name of that
branch. As it is likely that we will be introducing more short-hands to
write the name of a branch without writing it explicitly, rename the
functions from "nth_last_branch" to more generic "branch_name", to prepare
for different semantics.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* mv/parseopt-ls-files:
ls-files: fix broken --no-empty-directory
t3000: use test_cmp instead of diff
parse-opt: migrate builtin-ls-files.
Turn the flags in struct dir_struct into a single variable
Conflicts:
builtin-ls-files.c
t/t3000-ls-files-others.sh
Code that calls diff_setup(), including via init_revisions(), should later call
diff_setup_done(), possibly via setup_revisions(). Failure to do so could cause
errors, especially in the future when we add responsibilities to
diff_setup_done(). This instance causes no known errors with the present code.
But it resulted in an error with an experimental patch.
Signed-off-by: Keith Cascio <keith@cs.ucla.edu>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By having flags represented as bits in the new member variable 'flags',
it will be easier to use parse_options when dir_struct is involved.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Have '-' mean the same as '@{-1}', i.e., the last branch we were on.
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Have the lookup only look at "interesting" checkouts, meaning those
that tell you "Already on ..." don't count even though they also cause
a reflog entry.
Let interpret_nth_last_branch() return the number of characters
parsed, so that git-checkout can verify that the branch spec was
@{-N}, not @{-1}^2 or something like that. (The latter will be added
later.)
Signed-off-by: Thomas Rast <trast@student.ethz.ch>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
