summaryrefslogtreecommitdiff
path: root/builtin
Commit message (Collapse)AuthorAgeFilesLines
* worktree: usage: denote <branch> as optional with 'add'es/worktree-addSidhant Sharma2015-10-181-1/+1
| | | | | | | | | | | | Although 1eb07d8 (worktree: add: auto-vivify new branch when <branch> is omitted, 2015-07-06) updated the documentation when <branch> became optional, it neglected to update the in-code usage message. Fix this oversight. Reported-by: ch3cooli@gmail.com Signed-off-by: Sidhant Sharma <tigerkid001@gmail.com> Helped-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* config: rename "gc.pruneWorktreesExpire" to "gc.worktreePruneExpire"Eric Sunshine2015-07-201-1/+1
| | | | | | | | | | As of df0b6cf (worktree: new place for "git prune --worktrees", 2015-06-29), linked worktree pruning functionality moved from "git prune --worktrees" to "git worktree prune". Rename the associated configuration variable accordingly. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* Revert "checkout: retire --ignore-other-worktrees in favor of --force"Junio C Hamano2015-07-122-4/+6
| | | | | | | | | | | | This reverts commit 0d1a1517835a10818f2d40d8780a268dbb5e20ce. When trying to switch to a different branch, that happens to be checked out in another working tree, the user shouldn't have to give up the other safety measures (like protecting the local changes that overlap the difference between the branches) while defeating the "no two checkouts of the same branch" safety. Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: retire --ignore-other-worktrees in favor of --forceEric Sunshine2015-07-072-6/+4
| | | | | | | | | | | | As a safeguard, checking out a branch already checked out by a different worktree is disallowed. This behavior can be overridden with --ignore-other-worktrees, however, this option is neither obvious nor particularly discoverable. As a common safeguard override, --force is more likely to come to mind. Therefore, overload it to also suppress the check for a branch already checked out elsewhere. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: add: auto-vivify new branch when <branch> is omittedEric Sunshine2015-07-071-2/+6
| | | | | | | | | | | As a convenience, when <branch> is omitted from "git worktree <path> <branch>" and neither -b nor -B is used, automatically create a new branch named after <path>, as if "-b $(basename <path>)" was specified. Thus, "git worktree add ../hotfix" creates a new branch named "hotfix" and associates it with new worktree "../hotfix". Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: add: make -b/-B default to HEAD when <branch> is omittedEric Sunshine2015-07-061-2/+4
| | | | | | | | | As a convenience, like "git branch" and "git checkout -b", make "git worktree add -b <newbranch> <path> <branch>" default to HEAD when <branch> is omitted. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: extract basename computation to new functionEric Sunshine2015-07-061-9/+20
| | | | | | | | A subsequent patch will also need to compute the basename of the new worktree, so factor out this logic into a new function. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: require worktree unconditionallyEric Sunshine2015-07-061-2/+0
| | | | | | | | | | | | | | | | | In order to allow linked worktree creation via "git checkout --to" from a bare repository, 3473ad0 (checkout: don't require a work tree when checking out into a new one, 2014-11-30) dropped git-checkout's unconditional NEED_WORK_TREE requirement and instead performed worktree setup conditionally based upon presence or absence of the --to option. Now that --to has been retired and git-checkout is no longer responsible for linked worktree creation, the NEED_WORK_TREE requirement can be re-instated. This effectively reverts 3473ad0, except for the tests it added which now check bare repository behavior of "git worktree add" instead. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: retire --to optionEric Sunshine2015-07-062-166/+139
| | | | | | | | | | | | | | | | | | | Now that "git worktree add" has achieved user-facing feature-parity with "git checkout --to", retire the latter. Move the actual linked worktree creation functionality, prepare_linked_checkout() and its helpers, verbatim from checkout.c to worktree.c. This effectively reverts changes to checkout.c by 529fef2 (checkout: support checking out into a new working directory, 2014-11-30) with the exception of merge_working_tree() and switch_branches() which still require specialized knowledge that a the checkout is occurring in a newly-created linked worktree (signaled to them by the private GIT_CHECKOUT_NEW_WORKTREE environment variable). Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: add -b/-B optionsEric Sunshine2015-07-061-0/+11
| | | | | | | | | | | | One of git-worktree's roles is to populate the new worktree, much like git-checkout, and thus, for convenience, ought to support several of the same shortcuts. Toward this goal, add -b/-B options to create a new branch and check it out in the new worktree. (For brevity, only -b is mentioned in the synopsis; -B is omitted.) Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: add --detach optionEric Sunshine2015-07-061-1/+4
| | | | | | | | | | One of git-worktree's roles is to populate the new worktree, much like git-checkout, and thus, for convenience, ought to support several of the same shortcuts. Toward this goal, add a --detach option to detach HEAD in the new worktree. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: add --force optionEric Sunshine2015-07-061-1/+5
| | | | | | | | | By default, "git worktree add" refuses to create a new worktree when the requested branch is already checked out elsewhere. Add a --force option to override this safeguard. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: introduce "add" commandEric Sunshine2015-07-061-0/+31
| | | | | | | | | | | | The plan is to relocate "git checkout --to" functionality to "git worktree add". As a first step, introduce a bare-bones git-worktree "add" command along with documentation. At this stage, "git worktree add" merely invokes "git checkout --to" behind the scenes, but an upcoming patch will move the actual functionality (checkout.c:prepare_linked_checkout() and its helpers) to worktree.c. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: drop 'checkout_opts' dependency from prepare_linked_checkoutEric Sunshine2015-07-061-4/+5
| | | | | | | | | | | | | | The plan is to relocate "git checkout --to" functionality to "git worktree add", however, worktree.c won't have access to the 'struct checkout_opts' passed to prepare_linked_worktree(), which it consults for the pathname of the new worktree and the argv[] of the command it should run to populate the new worktree. Facilitate relocation of prepare_linked_worktree() by instead having it accept the pathname and argv[] directly, thus eliminating the final references to 'struct checkout_opts'. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: make --to unconditionally verboseEric Sunshine2015-07-061-2/+1
| | | | | | | | | | | | | | prepare_linked_checkout() respects git-checkout's --quiet flag, however, the plan is to relocate "git checkout --to" functionality to "git worktree add", and git-worktree does not (yet) have a --quiet flag. Consequently, make prepare_linked_checkout() unconditionally verbose to ease eventual code movement to worktree.c. (A --quiet flag can be added to git-worktree later if there is demand for it.) Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: prepare_linked_checkout: drop now-unused 'new' argumentEric Sunshine2015-07-061-3/+2
| | | | | | | The only references to 'new' were folded out by the last two patches. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: relocate --to's "no branch specified" checkEric Sunshine2015-07-061-3/+4
| | | | | | | | | | | | The plan is to relocate "git checkout --to" functionality to "git worktree add", however, this check expects a 'struct branch_info' which git-worktree won't have at hand. It will, however, have access to its own command-line from which it can pick up the branch name. Therefore, as a preparatory step, rather than having prepare_linked_checkout() perform this check, make it the caller's responsibility. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: fix bug with --to and relative HEADEric Sunshine2015-07-061-4/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Given "git checkout --to <path> HEAD~1", the new worktree's HEAD should begin life at the current branch's HEAD~1, however, it actually ends up at HEAD~2. This happens because: 1. git-checkout resolves HEAD~1 2. to satisfy is_git_directory(), prepare_linked_worktree() creates a HEAD for the new worktree with the value of the resolved HEAD~1 3. git-checkout re-invokes itself with the same arguments within the new worktree to populate the worktree 4. the sub git-checkout resolves HEAD~1 relative to its own HEAD, which is the resolved HEAD~1 from the original invocation, resulting unexpectedly and incorrectly in HEAD~2 (relative to the original) Fix this by unconditionally assigning the current worktree's HEAD as the value of the new worktree's HEAD. As a side-effect, this change also eliminates a dependence within prepare_linked_checkout() upon 'struct branch_info'. The plan is to eventually relocate "git checkout --to" functionality to "git worktree add", and worktree.c won't have knowledge of 'struct branch_info', so removal of this dependency is a step toward that goal. Signed-off-by: Eric Sunshine <sunshine@sunshineco.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* worktree: new place for "git prune --worktrees"nd/multiple-work-treesNguyễn Thái Ngọc Duy2015-06-293-100/+134
| | | | | | | | | Commit 23af91d (prune: strategies for linked checkouts - 2014-11-30) adds "--worktrees" to "git prune" without realizing that "git prune" is for object database only. This patch moves the same functionality to a new command "git worktree". Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
* checkout: don't check worktrees when not necessaryNguyễn Thái Ngọc Duy2015-06-121-12/+11
| | | | | | | | | | When --patch or pathspecs are passed to git checkout, the working tree will not be switching branch, so there's no need to check if the branch that we are running checkout on is already checked out. Original-patch-by: Spencer Baugh <sbaugh@catern.com> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* prune --worktrees: fix expire vs worktree existence conditionMax Kirillov2015-03-311-3/+7
| | | | | | | | | | | | | | | | | `git prune --worktrees` was pruning worktrees which were non-existent OR expired, while it rather should prune those which are orphaned AND expired, as git-checkout documentation describes. Fix it. Add test 'not prune proper checkouts', which uses valid but expired worktree. Modify test 'not prune recent checkouts' to remove the worktree before pruning - link in worktrees still must survive. In older form it is useless because would pass always when the other test passes. Signed-off-by: Max Kirillov <max@max630.net> Acked-by: Duy Nguyen <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: add --ignore-other-wortreesNguyễn Thái Ngọc Duy2015-01-071-1/+5
| | | | | | Noticed-by: Mark Levedahl <mlevedahl@gmail.com> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: pass whole struct to parse_branchname_arg instead of individual flagsNguyễn Thái Ngọc Duy2015-01-071-7/+6
| | | | | Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: do not fail if target is an empty directoryMax Kirillov2014-12-011-1/+1
| | | | | | | | | | | | | Non-recursive checkout creates empty directpries in place of submodules. If then I try to "checkout --to" submodules there, it refuses to do so, because directory already exists. Fix by allowing checking out to empty directory. Add test and modify the existing one so that it uses non-empty directory. Signed-off-by: Max Kirillov <max@max630.net> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: don't require a work tree when checking out into a new oneDennis Kaarsemaker2014-12-011-0/+3
| | | | | | | | | | | | | | | | | | | | | | | | For normal use cases, it does not make sense for 'checkout' to work on a bare repository, without a worktree. But "checkout --to" is an exception because it _creates_ a new worktree. Allow this option to run on bare repositories. People who check out from a bare repository should remember that core.logallrefupdates is off by default and it should be turned back on. `--to` cannot do this automatically behind the user's back because some user may deliberately want no reflog. For people interested in repository setup/discovery code, is_bare_repository_cfg (aka "core.bare") is unchanged by this patch, which means 'true' by default for bare repos. Fortunately when we get the repo through a linked checkout, is_bare_repository_cfg is never used. So all is still good. [nd: commit message] Signed-off-by: Dennis Kaarsemaker <dennis@kaarsemaker.net> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* count-objects: report unused files in $GIT_DIR/worktrees/...Nguyễn Thái Ngọc Duy2014-12-011-1/+3
| | | | | | | | | In linked checkouts, borrowed parts like config is taken from $GIT_COMMON_DIR. $GIT_DIR/config is never used. Report them as garbage. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* gc: support prune --worktreesNguyễn Thái Ngọc Duy2014-12-011-0/+10
| | | | | | | Helped-by: Marc Branchaud <marcnarc@xiplink.com> Signed-off-by: Marc Branchaud <marcnarc@xiplink.com> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* gc: factor out gc.pruneexpire parsing codeNguyễn Thái Ngọc Duy2014-12-011-10/+12
| | | | | Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* gc: style change -- no SP before closing parenthesisNguyễn Thái Ngọc Duy2014-12-011-1/+1
| | | | | Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: clean up half-prepared directories in --to modeNguyễn Thái Ngọc Duy2014-12-011-0/+48
| | | | | Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: reject if the branch is already checked out elsewhereNguyễn Thái Ngọc Duy2014-12-011-2/+84
| | | | | | | | | | One branch obviously can't be checked out at two places (but detached heads are ok). Give the user a choice in this case: --detach, -b new-branch, switch branch in the other checkout first or simply 'cd' and continue to work there. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* prune: strategies for linked checkoutsNguyễn Thái Ngọc Duy2014-12-012-2/+112
| | | | | | | | | | | | | | | | | | | | | | | | | | | (alias R=$GIT_COMMON_DIR/worktrees/<id>) - linked checkouts are supposed to keep its location in $R/gitdir up to date. The use case is auto fixup after a manual checkout move. - linked checkouts are supposed to update mtime of $R/gitdir. If $R/gitdir's mtime is older than a limit, and it points to nowhere, worktrees/<id> is to be pruned. - If $R/locked exists, worktrees/<id> is not supposed to be pruned. If $R/locked exists and $R/gitdir's mtime is older than a really long limit, warn about old unused repo. - "git checkout --to" is supposed to make a hard link named $R/link pointing to the .git file on supported file systems to help detect the user manually deleting the checkout. If $R/link exists and its link count is greated than 1, the repo is kept. Helped-by: Marc Branchaud <marcnarc@xiplink.com> Helped-by: Eric Sunshine <sunshine@sunshineco.com> Helped-by: Johannes Sixt <j6t@kdbg.org> Signed-off-by: Marc Branchaud <marcnarc@xiplink.com> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* checkout: support checking out into a new working directoryNguyễn Thái Ngọc Duy2014-12-011-2/+93
| | | | | | | | | | | | | "git checkout --to" sets up a new working directory with a .git file pointing to $GIT_DIR/worktrees/<id>. It then executes "git checkout" again on the new worktree with the same arguments except "--to" is taken out. The second checkout execution, which is not contaminated with any info from the current repository, will actually check out and everything that normal "git checkout" does. Helped-by: Marc Branchaud <marcnarc@xiplink.com> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* use new wrapper write_file() for simple file writingNguyễn Thái Ngọc Duy2014-12-012-9/+2
| | | | | | | | | This fixes common problems in these code about error handling, forgetting to close the file handle after fprintf() fails, or not printing out the error string.. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* setup.c: support multi-checkout repo setupNguyễn Thái Ngọc Duy2014-12-011-0/+4
| | | | | | | | | | | | | | The repo setup procedure is updated to detect $GIT_DIR/commondir and set $GIT_COMMON_DIR properly. The core.worktree is ignored when $GIT_COMMON_DIR is set. This is because the config file is shared in multi-checkout setup, but checkout directories _are_ different. Making core.worktree effective in all checkouts mean it's back to a single checkout. Helped-by: Johannes Sixt <j6t@kdbg.org> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* commit: use SEQ_DIR instead of hardcoding "sequencer"Nguyễn Thái Ngọc Duy2014-12-011-1/+1
| | | | | Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* reflog: avoid constructing .lock path with git_pathNguyễn Thái Ngọc Duy2014-12-011-1/+1
| | | | | | | | | | | | | | | | | | | Among pathnames in $GIT_DIR, e.g. "index" or "packed-refs", we want to automatically and silently map some of them to the $GIT_DIR of the repository we are borrowing from via $GIT_COMMON_DIR mechanism. When we formulate the pathname for its lockfile, we want it to be in the same location as its final destination. "index" is not shared and needs to remain in the borrowing repository, while "packed-refs" is shared and needs to go to the borrowed repository. git_path() could be taught about the ".lock" suffix and map "index.lock" and "packed-refs.lock" the same way their basenames are mapped, but instead the caller can help by asking where the basename (e.g. "index") is mapped to git_path() and then appending ".lock" after the mapping is done. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* git_path(): be aware of file relocation in $GIT_DIRNguyễn Thái Ngọc Duy2014-12-011-0/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | We allow the user to relocate certain paths out of $GIT_DIR via environment variables, e.g. GIT_OBJECT_DIRECTORY, GIT_INDEX_FILE and GIT_GRAFT_FILE. Callers are not supposed to use git_path() or git_pathdup() to get those paths. Instead they must use get_object_directory(), get_index_file() and get_graft_file() respectively. This is inconvenient and could be missed in review (for example, there's git_path("objects/info/alternates") somewhere in sha1_file.c). This patch makes git_path() and git_pathdup() understand those environment variables. So if you set GIT_OBJECT_DIRECTORY to /foo/bar, git_path("objects/abc") should return /foo/bar/abc. The same is done for the two remaining env variables. "git rev-parse --git-path" is the wrapper for script use. This patch kinda reverts a0279e1 (setup_git_env: use git_pathdup instead of xmalloc + sprintf - 2014-06-19) because using git_pathdup here would result in infinite recursion: setup_git_env() -> git_pathdup("objects") -> .. -> adjust_git_path() -> get_object_directory() -> oops, git_object_directory is NOT set yet -> setup_git_env() I wanted to make git_pathdup_literal() that skips adjust_git_path(). But that won't work because later on when $GIT_COMMON_DIR is introduced, git_pathdup_literal("objects") needs adjust_git_path() to replace $GIT_DIR with $GIT_COMMON_DIR. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* git_snpath(): retire and replace with strbuf_git_path()Nguyễn Thái Ngọc Duy2014-12-011-5/+8
| | | | | | | | | | | | | | | | In the previous patch, git_snpath() is modified to allocate a new strbuf buffer because vsnpath() needs that. But that makes it awkward because git_snpath() receives a pre-allocated buffer from outside and has to copy data back. Rename it to strbuf_git_path() and make it receive strbuf directly. Using git_path() in update_refs_for_switch() which used to call git_snpath() is safe because that function and all of its callers do not keep any pointer to the round-robin buffer pool allocated by get_pathname(). Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* path.c: make get_pathname() call sites return const char *Nguyễn Thái Ngọc Duy2014-12-017-14/+18
| | | | | | | | | | | | | | | | | | | Before the previous commit, get_pathname returns an array of PATH_MAX length. Even if git_path() and similar functions does not use the whole array, git_path() caller can, in theory. After the commit, get_pathname() may return a buffer that has just enough room for the returned string and git_path() caller should never write beyond that. Make git_path(), mkpath() and git_path_submodule() return a const buffer to make sure callers do not write in it at all. This could have been part of the previous commit, but the "const" conversion is too much distraction from the core changes in path.c. Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* Merge branch 'jk/fetch-reflog-df-conflict'Junio C Hamano2014-11-061-1/+1
|\ | | | | | | | | | | | | | | Corner-case bugfixes for "git fetch" around reflog handling. * jk/fetch-reflog-df-conflict: ignore stale directories when checking reflog existence fetch: load all default config at startup
| * fetch: load all default config at startupJeff King2014-11-041-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When we start the git-fetch program, we call git_config to load all config, but our callback only processes the fetch.prune option; we do not chain to git_default_config at all. This means that we may not load some core configuration which will have an effect. For instance, we do not load core.logAllRefUpdates, which impacts whether or not we create reflogs in a bare repository. Note that I said "may" above. It gets even more exciting. If we have to transfer actual objects as part of the fetch, then we call fetch_pack as part of the same process. That function loads its own config, which does chain to git_default_config, impacting global variables which are used by the rest of fetch. But if the fetch is a pure ref update (e.g., a new ref which is a copy of an old one), we skip fetch_pack entirely. So we get inconsistent results depending on whether or not we have actual objects to transfer or not! Let's just load the core config at the start of fetch, so we know we have it (we may also load it again as part of fetch_pack, but that's OK; it's designed to be idempotent). Our tests check both cases (with and without a pack). We also check similar behavior for push for good measure, but it already works as expected. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* | Merge branch 'jc/push-cert'Junio C Hamano2014-10-311-2/+2
|\ \ | | | | | | | | | | | | * jc/push-cert: receive-pack: avoid minor leak in case start_async() fails
| * | receive-pack: avoid minor leak in case start_async() failsRené Scharfe2014-10-281-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | If the asynchronous start of copy_to_sideband() fails, then any env_array entries added to struct child_process proc by prepare_push_cert_sha1() are leaked. Call the latter function only after start_async() succeeded so that the allocated entries are cleaned up automatically by start_command() or finish_command(). Signed-off-by: Rene Scharfe <l.s.r@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* | | Merge branch 'jk/prune-mtime'Junio C Hamano2014-10-297-199/+157
|\ \ \ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Tighten the logic to decide that an unreachable cruft is sufficiently old by covering corner cases such as an ancient object becoming reachable and then going unreachable again, in which case its retention period should be prolonged. * jk/prune-mtime: (28 commits) drop add_object_array_with_mode revision: remove definition of unused 'add_object' function pack-objects: double-check options before discarding objects repack: pack objects mentioned by the index pack-objects: use argv_array reachable: use revision machinery's --indexed-objects code rev-list: add --indexed-objects option rev-list: document --reflog option t5516: test pushing a tag of an otherwise unreferenced blob traverse_commit_list: support pending blobs/trees with paths make add_object_array_with_context interface more sane write_sha1_file: freshen existing objects pack-objects: match prune logic for discarding objects pack-objects: refactor unpack-unreachable expiration check prune: keep objects reachable from recent objects sha1_file: add for_each iterators for loose and packed objects count-objects: use for_each_loose_file_in_objdir count-objects: do not use xsize_t when counting object size prune-packed: use for_each_loose_file_in_objdir reachable: mark index blobs as SEEN ...
| * | | pack-objects: double-check options before discarding objectsJeff King2014-10-191-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When we are given an expiration time like --unpack-unreachable=2.weeks.ago, we avoid writing out old, unreachable loose objects entirely, under the assumption that running "prune" would simply delete them immediately anyway. However, this is only valid if we computed the same set of reachable objects as prune would. In practice, this is the case, because only git-repack uses the --unpack-unreachable option with an expiration, and it always feeds as many objects into the pack as possible. But we can double-check at runtime just to be sure. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
| * | | repack: pack objects mentioned by the indexJeff King2014-10-192-0/+9
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When we pack all objects, we use only the objects reachable from references and reflogs. This misses any objects which are reachable from the index, but not yet referenced. By itself this isn't a big deal; the objects can remain loose until they are actually used in a commit. However, it does create a problem when we drop packed but unreachable objects. We try to optimize out the writing of objects that we will immediately prune, which means we must follow the same rules as prune in determining what is reachable. And prune uses the index for this purpose. This is rather uncommon in practice, as objects in the index would not usually have been packed in the first place. But it could happen in a sequence like: 1. You make a commit on a branch that references blob X. 2. You repack, moving X into the pack. 3. You delete the branch (and its reflog), so that X is unreferenced. 4. You "git add" blob X so that it is now referenced only by the index. 5. You repack again with git-gc. The pack-objects we invoke will see that X is neither referenced nor recent and not bother loosening it. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
| * | | pack-objects: use argv_arrayJeff King2014-10-191-10/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This saves us from having to bump the rp_av count when we add new traversal options. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
| * | | make add_object_array_with_context interface more saneJeff King2014-10-161-4/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When you resolve a sha1, you can optionally keep any context found during the resolution, including the path and mode of a tree entry (e.g., when looking up "HEAD:subdir/file.c"). The add_object_array_with_context function lets you then attach that context to an entry in a list. Unfortunately, the interface for doing so is horrible. The object_context structure is large and most object_array users do not use it. Therefore we keep a pointer to the structure to avoid burdening other users too much. But that means when we do use it that we must allocate the struct ourselves. And the struct contains a fixed PATH_MAX-sized buffer, which makes this wholly unsuitable for any large arrays. We can observe that there is only a single user of the "with_context" variant: builtin/grep.c. And in that use case, the only element we care about is the path. We can therefore store only the path as a pointer (the context's mode field was redundant with the object_array_entry itself, and nobody actually cared about the surrounding tree). This still requires a strdup of the pathname, but at least we are only consuming the minimum amount of memory for each string. We can also handle the copying ourselves in add_object_array_*, and free it as appropriate in object_array_release_entry. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
| * | | pack-objects: match prune logic for discarding objectsJeff King2014-10-161-0/+39
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | A recent commit taught git-prune to keep non-recent objects that are reachable from recent ones. However, pack-objects, when loosening unreachable objects, tries to optimize out the write in the case that the object will be immediately pruned. It now gets this wrong, since its rule does not reflect the new prune code (and this can be seen by running t6501 with a strategically placed repack). Let's teach pack-objects similar logic. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
View Lorry Controller Status