* wc/rebase-insn:
Mention that git-rm can be an appropriate resolution as well as git-add.
revert/cherry-pick: Allow overriding the help text by the calling Porcelain
git-stash: Display help message if git-stash is run with wrong sub-commands
The current git-stash behaviour is very error prone to typos. For example,
if you typed "git-stash llist", git-stash would think that you wanted to
save to a stash named "llist", but in fact, you meant "git-stash list".
Signed-off-by: Kevin Leung <kevinlsk@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Even though commit-tree would default to the current time if the incoming
e-mail message somehow did not record the timestamp, it is safer to catch
the breakage sooner.
When we consider if a path has been totally rewritten, we did not
touch changes from symlinks to files or vice versa. But a change
that modifies even the type of a blob surely should count as a
complete rewrite.
While we are at it, modernise diffcore-break to be aware of gitlinks (we
do not want to touch them).
The categorized list of commands in git(7) and the list of common
commands in "git help" output were maintained separately, which was
insane. This consolidates them to a single command-list.txt file.
8951d7c1f1ae38f34617b6c2490bf65e73e371f7 (Build in ls-remote) made
peek-remote as a synonym to ls-remote by enhancing the latter, but
at the same time actually _removed_ it, before we officially gave
removal notice. This was bad.
* cr/tag-options:
git-tag: test that -s implies an annotated tag
"git-tag -s" should create a signed annotated tag
builtin-tag: accept and process multiple -m just like git-commit
Make builtin-tag.c use parse_options.
git-svn: Don't create a "master" branch every time rebase is run
If you run "git-svn rebase" while sitting on a topic branch, there is
no need to create a "master" branch if one didn't exist already. The
branch was created implicitly by the automatic checkout after fetching,
which in the case of rebase isn't actually necessary anyway.
Signed-off-by: Steven Grimm <koreth@midwinter.com> Acked-by: Eric Wong <normalperson@yhbt.net>
show-externals can be used by scripts to provide svn:externals-like
functionality. For example, a script can list all of the externals and then
use check out the listed URLs at the appropriate paths, similar to what the svn
client does. Said script (or perhaps git-svn itself, in the future) could
simply invoke svn export on the paths, or it could go one further, using
git-svn clone and even git-submodule together to better integrate externals
checkouts.
The implementation is shamelessly copied from show-ignores. A more general
command to list user-specified properties is probably a better idea.
Signed-off-by: Vineet Kumar <vineet@doorstop.net> Acked-by: Eric Wong <normalperson@yhbt.net>
git-svn now reads settings even if called in subdirectory
Previously, git-svn first read the .git/config file for settings as if
current working directory was the repository top-directory, and after
that made sure to cd into top-directory. The result was a silent
failur to read configuration settings. This patch changes the order
these two things are done.
Signed-off-by: Gustaf Hendeby <hendeby@isy.liu.se> Acked-by: Eric Wong <normalperson@yhbt.net>
* maint:
Replace the word 'update-cache' by 'update-index' everywhere
cvsimport: fix usage of cvsimport.module
t7003-filter-branch: Fix test of a failing --msg-filter.
cvsimport: miscellaneous packed-ref fixes
cvsimport: use rev-parse to support packed refs
Add basic cvsimport tests
Fix a pathological case in git detecting proper renames
On Thu, 29 Nov 2007, Jeff King wrote:
>
> I think it will get worse, because you are simultaneously calculating
> all of the similarity scores bit by bit rather than doing a loop. Though
> perhaps you mean at the end you will end up with a list of src/dst pairs
> sorted by score, and you can loop over that.
Well, after thinking about this a bit, I think there's a solution that may
work well with the current thing too: instead of looping just *once* over
the list of rename pairs, loop twice - and simply refuse to do copies on
the first loop.
This trivial patch does that, and turns Kumar's test-case into a perfect
rename list.
It's not pretty, it's not smart, but it seems to work. There's something
to be said for keeping it simple and stupid.
And it should not be nearly as expensive as it may _look_. Yes, the loop
is "(i = 0; i < num_create * num_src; i++)", but the important part is
that the whole array is sorted by rename score, and we have a
if (mx[i].score < minimum_score)
break;
in it, so uthe loop actually would tend to terminate rather quickly.
Anyway, Kumar, the thing to take away from this is:
- git really doesn't even *care* about the whole "rename detection"
internally, and any commits you have done with renames are totally
independent of the heuristics we then use to *show* the renames.
- the rename detection really is for just two reasons: (a) keep humans
happy, and keep the diffs small and (b) help automatic merging across
renames. So getting renames right is certainly good, but it's more of a
"politeness" issue than a "correctness" issue, although the merge
portion of it does matter a lot sometimes.
- the important thing here is that you can commit your changes and not
worry about them being somehow "corrupted" by lack of rename detection,
even if you commit them with a version of git that doesn't do rename
detection the way you expected it. The rename detection is an
"after-the-fact" thing, not something that actually gets saved in the
repository, which is why we can change the heuristics _after_ seeing
examples, and the examples magically correct themselves!
- try out the two patches I've posted, and see if they work for you. They
pass the test-suite, and the output for your example commit looks sane,
but hey, if you have other test-cases, try them out.
Here's Kumar's pretty diffstat with both my patches:
Fix a pathological case in git detecting proper renames
Kumar Gala had a case in the u-boot archive with multiple renames of files
with identical contents, and git would turn those into multiple "copy"
operations of one of the sources, and just deleting the other sources.
This patch makes the git exact rename detection prefer to spread out the
renames over the multiple sources, rather than do multiple copies of one
source.
NOTE! The changes are a bit larger than required, because I also renamed
the variables named "one" and "two" to "target" and "source" respectively.
That makes the logic easier to follow, especially as the "one" was
illogically the target and not the soruce, for purely historical reasons
(this piece of code used to traverse over sources and targets in the wrong
order, and when we fixed that, we didn't fix the names back then. So I
fixed them now).
The important part of this change is just the trivial score calculations
for when files have identical contents:
/* Give higher scores to sources that haven't been used already */
score = !source->rename_used;
score += basename_same(source, target);
and when we have multiple choices we'll now pick the choice that gets the
best rename score, rather than only looking at whether the basename
matched.
It's worth noting a few gotchas:
- this scoring is currently only done for the "exact match" case.
In particular, in Kumar's example, even after this patch, the inexact
match case is still done as a copy+delete rather than as two renames:
because apparently the "cds/mpc8541cds/u-boot.lds" copy looked
a bit more similar to both end results. That said, I *suspect* we just
have the exact same issue there - the similarity analysis just gave
identical (or at least very _close_ to identical) similarity points,
and we do not have any logic to prefer multiple renames over a
copy/delete there.
That is a separate patch.
- When you have identical contents and identical basenames, the actual
entry that is chosen is still picked fairly "at random" for the first
one (but the subsequent ones will prefer entries that haven't already
been used).
It's not actually really random, in that it actually depends on the
relative alphabetical order of the files (which in turn will have
impacted the order that the entries got hashed!), so it gives
consistent results that can be explained. But I wanted to point it out
as an issue for when anybody actually does cross-renames.
In Kumar's case the choice is the right one (and for a single normal
directory rename it should always be, since the relative alphabetical
sorting of the files will be identical), and we now get:
which is the "expected" answer. However, it might still be better to
change the pedantic "exact same basename" on/off choice into a more
graduated "how similar are the pathnames" scoring situation, in order
to be more likely to get the exact rename choice that people *expect*
to see, rather than other alternatives that may *technically* be
equally good, but are surprising to a human.
It's also unclear whether we should consider "basenames are equal" or
"have already used this as a source" to be more important. This gives them
equal weight, but I suspect we might want to just multiple the "basenames
are equal" weight by two, or something, to prefer equal basenames even if
that causes a copy/delete pair. I dunno.
Anyway, what I'm just saying in a really long-winded manner is that I
think this is right as-is, but it's not the complete solution, and it may
want some further tweaking in the future.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
1. We only look at the config variable if there is no module
given on the command line. We checked this by comparing
@ARGV == 0. However, at the time of the comparison, we
have not yet parsed the dashed options, meaning that
"git cvsimport" would read the variable but "git
cvsimport -a" would not. This is fixed by simply moving
the check after the call to getopt.
2. If the config variable did not exist, we were adding an
empty string to @ARGV. The rest of the script, rather
than barfing for insufficient input, would then try to
import the module '', leading to rather confusing error
messages. Based on patch from Emanuele Giaquinta.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Occasionally, in some setups (*cough* forks on repo.or.cz *cough*) some
refs go stale, e.g. when the forkee rebased and lost some objects needed
by the fork. The quick & dirty way to deal with those refs is to delete
them and push them again.
However, git-push first would first fetch the current commit name for the
ref, would receive a null sha1 since the ref does not point to a valid
object, then tell receive-pack that it should delete the ref with this
commit name. delete_ref() would be subsequently be called, and check that
resolve_ref() (which does _not_ check for validity of the object) returns
the same commit name. Which would fail.
The proper fix is to avoid corrupting repositories, but in the meantime
this is a good fix in any case.
Incidentally, some instances of "cd .." in the test cases were fixed, so
that subsequent test cases run in t/trash/ irrespective of the outcome of
the previous test cases.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
t7003-filter-branch: Fix test of a failing --msg-filter.
The test passed for the wrong reason: If the script given to --msg-filter
fails, it is expected that git-filter-branch aborts. But the test forgot
to tell the branch name to rewrite, and so git-filter-branch failed due to
incorrect usage.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Mention that git-rm can be an appropriate resolution as well as git-add.
Especially when using git-cherry-pick, removing files that are unmerged can be
a logical action. This patch merely changes the informative text to be less
confusing.
Signed-off-by: David Symonds <dsymonds@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
revert/cherry-pick: Allow overriding the help text by the calling Porcelain
A Porcelain command that uses cherry-pick or revert may make a commit
out of resolved index itself, in which case telling the user to commit
the result is not appropriate at all. This allows GIT_CHERRY_PICK_HELP
environment variable to be set by the calling Porcelain in order to
override the built-in help text.
[jc: this is heavily modified from the original but should be equivalent
in spirit]
Operations that walk directories or trees, which potentially need to
consult the .gitignore files, used to always try to open the .gitignore
file every time they entered a new directory, even when they ended up
not needing to call excluded() function to see if a path in the
directory is ignored. This was done by push/pop exclude_per_directory()
functions that managed the data in a stack.
This changes the directory walking API to remove the need to call these
two functions. Instead, the directory walk data structure caches the
data used by excluded() function the last time, and lazily reuses it as
much as possible. Among the data the last check used, the ones from
deeper directories that the path we are checking is outside are
discarded, data from the common leading directories are reused, and then
the directories between the common directory and the directory the path
being checked is in are checked for .gitignore file. This is very
similar to the way gitattributes are handled.
This API change also fixes "ls-files -c -i", which called excluded()
without setting up the gitignore data via the old push/pop functions.
Too many people got burned by setting color.diff and color.status to
true when they really should have set it to "auto".
This makes only "always" to do the unconditional colorization, and
change the meaning of "true" to the same as "auto": colorize only when
we are talking to a terminal.
When calling 'git pull' with the '--rebase' option, it performs a
fetch + rebase instead of a fetch + merge.
This behavior is more desirable than fetch + pull when a topic branch
is ready to be submitted and needs to be update.
fetch + rebase might also be considered a better workflow with shared
repositories in any case, or for contributors to a centrally managed
repository, such as WINE's.
As a convenience, you can set the default behavior for a branch by
defining the config variable branch.<name>.rebase, which is
interpreted as a bool. This setting can be overridden on the command
line by --rebase and --no-rebase.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
scripts: do not get confused with HEAD in work tree
When you have a file called HEAD in your work tree, many commands that
our scripts feed "HEAD" to would complain about the rev vs path
ambiguity. A solution is to form command line more carefully by
appending -- to them, which makes it clear that we mean HEAD rev not
HEAD file.
I use "git diff" (the porcelain) really often, and am almost as often
annoyed that the completions do not know how to complete something simple
as --cached. Now they do.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
sha1_file.c: Fix size_t related printf format warnings
The old way of fixing warnings did not succeed on MinGW. MinGW
does not support C99 printf format strings for size_t [1]. But
gcc on MinGW issues warnings if C99 printf format is not used.
Hence, the old stragegy to avoid warnings fails.
[1] http://www.mingw.org/MinGWiki/index.php/C99
This commits passes arguments of type size_t through a tiny
helper functions that casts to the type expected by the format
string.
Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Improve description of git-branch -d and -D in man page.
Some users expect that deleting a remote-tracking branch would prevent
fetch from creating it again, so be explcit about that it's not the case.
Also be a little more explicit about what fully merged means.
Signed-off-by: Jan Hudec <bulb@ucw.cz> Signed-off-by: Junio C Hamano <gitster@pobox.com>
filter-branch: fix dirty way to provide the helpers to commit filters
The helper functions 'map' and 'skip_commit' were provided to commit
filters by sourcing filter-branch itself. This was done with a certain
environment variable set to indicate that only the functions should be
defined, and the script should return then.
This was really hacky, and it did not work all that well, since the
full path to git-filter-branch was not known at all times.
Avoid that by putting the functions into a variable, and eval'ing
that variable. The commit filter gets these functions by prepending
the variable to the specified commands.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously, if refs were packed, git-cvsimport would assume
that particular refs did not exist. This could lead to, for
example, overwriting previous 'origin' commits that were
packed.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-stash: do not get fooled with "color.diff = true"
When colors are set to "true" on the repository, the git log output
will contain control characters to set/reset the colors, even when
the output is to a pipe. This makes list_stash() fail as the
downstream sed does not see what it is expecting.
* maint:
user-manual: recovering from corruption
user-manual: clarify language about "modifying" old commits
user-manual: failed push to public repository
user-manual: define "branch" and "working tree" at start
git-checkout: describe detached head correctly
Merge branch 'maint' of git://linux-nfs.org/~bfields/git into maint
* 'maint' of git://linux-nfs.org/~bfields/git:
user-manual: recovering from corruption
user-manual: clarify language about "modifying" old commits
user-manual: failed push to public repository
user-manual: define "branch" and "working tree" at start
user-manual: clarify language about "modifying" old commits
It's important to remember that git doesn't really allowing "editing" or
"modifying" commits, only replacing them by new commits. Redo some of
the language to make this clearer.
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
revert/cherry-pick: do not mention the original ref
When you cherry-pick or revert a commit, naming it with an annotated
tag, we added a comment, attempting to repeat what we got from the end
user, to the message.
But this was inconsistent. When we got "cherry-pick branch", we
recorded the object name (40-letter SHA-1) without saying anything like
"original was 'branch'". There was no need to. Also recent rewrite to
use parse-options made it impossible to parrot the original command line
without "unparsing".
This removes the code that implements the misguided "we dereferenced the
tag so record that in the commit message" behaviour.
When you have a file called HEAD in the work tree, the code to report
where the HEAD is at when "git checkout $commit^0" is done triggered
unnecessary ambiguity checking.
Explicitly mark the command line with "--" and make it clear that we are
talking about a revision.
* maint:
Fix rev-list when showing objects involving submodules
test format-patch -s: make sure MIME content type is shown as needed
format-patch -s: add MIME encoding header if signer's name requires so
* jc/maint-add-sync-stat:
t2200: test more cases of "add -u"
git-add: make the entry stat-clean after re-adding the same contents
ce_match_stat, run_diff_files: use symbolic constants for readability
Merge branch 'jc/maint-format-patch-encoding' into maint
* jc/maint-format-patch-encoding:
test format-patch -s: make sure MIME content type is shown as needed
format-patch -s: add MIME encoding header if signer's name requires so
* bs/maint-commit-options:
git-commit: Add tests for invalid usage of -a/--interactive with paths
git-commit.sh: Fix usage checks regarding paths given when they do not make sense
* jk/send-pack: (24 commits)
send-pack: cluster ref status reporting
send-pack: fix "everything up-to-date" message
send-pack: tighten remote error reporting
make "find_ref_by_name" a public function
Fix warning about bitfield in struct ref
send-pack: assign remote errors to each ref
send-pack: check ref->status before updating tracking refs
send-pack: track errors for each ref
git-push: add documentation for the newly added --mirror mode
Add tests for git push'es mirror mode
Update the tracking references only if they were succesfully updated on remote
Add a test checking if send-pack updated local tracking branches correctly
git-push: plumb in --mirror mode
Teach send-pack a mirror mode
send-pack: segfault fix on forced push
Reteach builtin-ls-remote to understand remotes
send-pack: require --verbose to show update of tracking refs
receive-pack: don't mention successful updates
more terse push output
Build in ls-remote
...
* js/mingw-fallouts:
fetch-pack: Prepare for a side-band demultiplexer in a thread.
rehabilitate some t5302 tests on 32-bit off_t machines
Allow ETC_GITCONFIG to be a relative path.
Introduce git_etc_gitconfig() that encapsulates access of ETC_GITCONFIG.
Allow a relative builtin template directory.
Close files opened by lock_file() before unlinking.
builtin run_command: do not exit with -1.
Move #include <sys/select.h> and <sys/ioctl.h> to git-compat-util.h.
Use is_absolute_path() in sha1_file.c.
Skip t3902-quoted.sh if the file system does not support funny names.
t5302-pack-index: Skip tests of 64-bit offsets if necessary.
t7501-commit.sh: Not all seds understand option -i
t5300-pack-object.sh: Split the big verify-pack test into smaller parts.
* cc/bisect:
Bisect reset: do nothing when not bisecting.
Bisect: use "$GIT_DIR/BISECT_NAMES" to check if we are bisecting.
Bisect visualize: use "for-each-ref" to list all good refs.
git-bisect: modernize branch shuffling hack
git-bisect: use update-ref to mark good/bad commits
git-bisect: war on "sed"
Bisect reset: remove bisect refs that may have been packed.
If the worktree happened to have a file called HEAD, "diff-index --cached HEAD"
would complain about the ambiguity between revision and path. Avoid it by
using an explicit "--" for disambiguation.
This teaches git-branch to limit its listing to branches that
are descendants to the named commit.
When you are using many topic branches, you often would want to
see which branch already includes a commit, so that you know
which can and cannot be rewound without disrupting other people.
One thing that sometimes happens to me is:
* Somebody sends a patch that is a good maint material. I
apply it to 'maint':
$ git checkout maint
$ git am -3 -s obvious-fix.patch
* Then somebody else sends another patch that is possibly a
good maint material, but I'd want to cook it in 'next' to be
extra sure. I fork a topic from 'maint' and apply the patch:
* A minor typo is found in the "obvious-fix.patch".
The above happens without pushing the results out, so I can
freely recover from it by amending 'maint', as long as I do not
forget to rebase the topics that were forked previously.
With this patch, I can do this to find out which topic
branches already contain the faulty commit:
$ git branch --contains=maint^
xx/maint-fix-foo
so I can rebase the xx/maint-fix-foo branch before merging it
to 'next'.
parse-options: Allow to hide options from the default usage.
This is useful for backward-compatibility aliases, or very advanced command
line switches introduced for internal git usages and have no real use for a
user.
parse-options still shows them if the user asks for --help-all.
Signed-off-by: Pierre Habouzit <madcoder@debian.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
bundle create: keep symbolic refs' names instead of resolving them
When creating a bundle, symbolic refs used to be resolved to the
non-symbolic refs they point to before being written to the list
of contained refs. I.e. "git bundle create a1.bundle HEAD master"
would show something like