This changes 'git-pack-refs' to pack only tags by default.
Branches are meant to be updated, either by committing onto it
yourself or tracking remote branches, and packed entries can
become stale easily, but tags are usually "create once and live
forever" and benefit more from packing.
core.logallrefupdates create new log file only for branch heads.
It used to mean "create log file for any ref that is updated",
but now it creates new log files only for branch heads.
The old behaviour made this configuration less useful than
otherwise it would be; automatically creating log file for tags
is almost always not useful.
* master: (99 commits)
lock_ref_sha1_basic does not remove empty directories on BSD
git-push: .git/remotes/ file does not require SP after colon
git-mv: invalidate the removed path properly in cache-tree
Makefile: install and clean merge-recur, still.
GIT 1.4.3-rc1
gitweb: tree view: hash_base and hash are now context sensitive
git-diff -B output fix.
fetch: Reset remote refs list each time fetch_main is called
Remove -fPIC which was only needed for Git.xs
Fix approxidate() to understand 12:34 AM/PM are 00:34 and 12:34
git-diff -B output fix.
Make cvsexportcommit remove files.
diff --stat: ensure at least one '-' for deletions, and one '+' for additions
diff --stat=width[,name-width]: allow custom diffstat output width.
gitweb: History: blob and tree are first, then commitdiff, etc
gitweb: Remove redundant "commit" from history
http/ftp: optionally ask curl to not use EPSV command
gitweb: Don't use quotemeta on internally generated strings
gitweb: Add snapshot to shortlog
gitweb: Factor out gitweb_have_snapshot()
...
lock_ref_sha1_basic does not remove empty directories on BSD
lock_ref_sha1_basic relies on errno beeing set to EISDIR by the
call to read() in resolve_ref() to detect directories. But calling
read() on a directory under NetBSD returns EPERM, and even succeeds
for local filesystems on FreeBSD.
Signed-off-by: Dennis Stosberg <dennis@stosberg.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-push: .git/remotes/ file does not require SP after colon
Although most people would have one after colon if only for
readability, we never required it in git-parse-remote, so let's
not require one only in git-push.
The function repack_without_ref() passes a lock-file structure
on the stack to hold_lock_file_for_update(), which in turn
registers it to be cleaned up via atexit(). This is a big
no-no.
* jc/gitpm: (52 commits)
Remove -fPIC which was only needed for Git.xs
Git.pm: Kill Git.xs for now
Revert "Make it possible to set up libgit directly (instead of from the environment)"
Revert "Git.pm: Introduce fast get_object() method"
Revert "Convert git-annotate to use Git.pm"
Fix compilation with Sun CC
pass DESTDIR to the generated perl/Makefile
Eliminate Scalar::Util usage from private-Error.pm
Convert git-annotate to use Git.pm
Git.pm: Introduce fast get_object() method
Make it possible to set up libgit directly (instead of from the environment)
Work around sed and make interactions on the backslash at the end of line.
Git.pm: Introduce ident() and ident_person() methods
Convert git-send-email to use Git.pm
Git.pm: Add config() method
Use $GITPERLLIB instead of $RUNNING_GIT_TESTS and centralize @INC munging
INSTALL: a tip for running after building but without installing.
Perly Git: make sure we do test the freshly built one.
Git.pm: Don't #define around die
Git.xs: older perl do not know const char *
...
* jc/diff-stat:
diff --stat: ensure at least one '-' for deletions, and one '+' for additions
diff --stat=width[,name-width]: allow custom diffstat output width.
diff --stat: color output.
diff --stat: allow custom diffstat output width.
* lt/web:
gitweb: tree view: hash_base and hash are now context sensitive
gitweb: History: blob and tree are first, then commitdiff, etc
gitweb: Remove redundant "commit" from history
gitweb: Don't use quotemeta on internally generated strings
gitweb: Add snapshot to shortlog
gitweb: Factor out gitweb_have_snapshot()
gitweb: Remove redundant "commit" link from shortlog
gitweb: "alternate" starts with shade (i.e. 1)
gitweb: Add history and blame to git_difftree_body()
gitweb: Remove excessively redundant entries from git_difftree_body
Revert "gitweb: extend blame to show links to diff and previous"
gitweb: Quote filename in HTTP Content-Disposition: header
gitweb: Add git_url subroutine, and use it to quote full URLs
gitweb: Split validate_input into validate_pathname and validate_refname
gitweb: Use "return" instead of "return undef" for some subs
gitweb: Strip trailing slashes from $path in git_get_hash_by_path
gitweb: extend blame to show links to diff and previous
gitweb: Remove redundant "tree" link
gitweb: tree view: eliminate redundant "blob"
Now ref creation codepath in lock_ref_sha1() and friends notices
the directory/file conflict situation, we do not do this by hand
in git-branch anymore.
This implements deletion of a packed ref. Since it is a very
rare event to delete a ref compared to looking up, creating and
updating, this opts to remove the ref from the packed-ref file
instead of doing any of the filesystem based "negative ref" trick
to optimize the deletion path.
lock_ref_sha1(): do not sometimes error() and sometimes die().
This cleans up the error path in the function so it does not
die() itself sometimes while signalling an error with NULL some
other times which was inconsistent and confusing.
Once we read packed and loose refs, for_each_ref() and friends
kept using them even after write_ref_sha1() and delete_ref()
changed the refs. This adds invalidate_cached_refs() as a way
to flush the cached information.
ref locking: allow 'foo' when 'foo/bar' used to exist but not anymore.
It is normal to have .git/refs/heads/foo directory which is
empty after the last branch whose name starts with foo/ is
removed. Make sure we notice this case and allow creation of
branch foo by removing the empty directory.
gitweb: tree view: hash_base and hash are now context sensitive
In tree view, by default, hash_base is HEAD and hash is the
entry equivalent. Else the user had selected a hash_base or
hash, say by clicking on a revision or commit, in which case
those values are used.
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
The distinction between BASIC_ vs ALL_ is still kept, since it
is not Git.xs specific -- we could face the same issue when we
do other language bindings (e.g. Python).
Fix approxidate() to understand 12:34 AM/PM are 00:34 and 12:34
It just simplifies the whole thing to say
"hour = (hour % 12) + X"
where X is 12 for PM and 0 for AM.
It also fixes the "exact date" parsing, which didn't parse AM at all, and
as such would do the same "12:30 AM" means "12:30 24-hour-format" bug. Of
course, I hope that no exact dates use AM/PM anyway, but since we support
the PM format, let's just get it right.
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
diff --stat: ensure at least one '-' for deletions, and one '+' for additions
The number of '-' and '+' is still linear. The idea is that
scaled-length := floor(a * length + b) with the following constraints: if
length == 1, scaled-length == 1, and the combined length of plusses
and minusses should not be larger than the width by a small margin. Thus,
a + b == 1
and
a * max_plusses + b + a * max_minusses + b = width + 1
The solution is
a * x + b = ((width - 1) * (x - 1) + max_change - 1)
/ (max_change - 1)
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <junkio@cox.net>
Remove redundant "commit" from history -- it can be had
by clicking on the title of the commit. This commit
makes visualization consistent with shortlog, log, etc.
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
http/ftp: optionally ask curl to not use EPSV command
If http.noEPSV config variable is defined and true, or if
GIT_CURL_FTP_NO_EPSV environment variable is defined, disable using
of EPSV ftp command (PASV will be used instead). This is helpful with
some "poor" ftp servers which does not support EPSV mode.
Signed-off-by: Sasha Khapyorsky <sashak@voltaire.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Don't use quotemeta on internally generated strings
Do not use quotemeta on internally generated strings
such as filenames of snapshot, blobs, etc.
quotemeta quotes any characters not matching /A-Za-z_0-9/.
Which means that we get strings like this:
When displaying a list of rows (difftree, shortlog, etc),
the first entry is now printed shaded, i.e. alternate is
initialized to 1, as opposed to non-shaded (alternate
initialized to 0).
This solves the problem when there is only one row to
display -- it is displayed shaded to visually indicate that
it is "active", part of a "list", etc.
(Compare this to the trivial case of more than one entry,
where the rows have alternating shade, thus suggesting
being part of a "list" of "active" entries, etc.)
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Our approxidate cannot handle simple times like "5 PM yesterday", and to
fix that, we will need to add some logic for number handling. This just
splits that out into a function of its own (the same way the _real_ date
parsing works).
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
svnimport: add support for parsing From: lines for author
When commiting a non-signed off contribution you cannot just add
a Signed-off-by: from the author as they did not sign it off.
But if you then commit it, and necessarily sign it off yourself,
the change appears to be yours. In this case it is common to use
the following form:
Commentry
From: originator <email> Signed-of-by: me <my email>
Now that we have support for parsing Signed-off-by: for author
information it makes sense to handle From: as well. This patch
adds a new -F which will handle From: lines in the comments. It
may be used in combination with -S.
Signed-off-by: Andy Whitcroft <apw@shadowen.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Contributed bash completion support for core Git tools.
This is a set of bash completion routines for many of the
popular core Git tools. I wrote these routines from scratch
after reading the git-compl and git-compl-lib routines available
from the gitcompletion package at http://gitweb.hawaga.org.uk/
and found those to be lacking in functionality for some commands.
Consequently there may be some similarities but many differences.
Since these are completion routines only for tools shipped with
core Git and since bash is a popular shell on many of the native
core Git platforms (Linux, Mac OS X, Solaris, BSD) including these
routines as part of the stock package would probably be convienent
for many users.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
When creating branch c/d check that branch c does not already exists.
With packed refs, there may not be a ".git/refs/heads/c" file
when branch c exists. And currently in this case, there is no check
to prevent creation of branch c/d.
This should probably be rewritten in C and done after the ref lock
has been taken to make sure no race exists though.
This is mainly to make all test cases in "t3210-pack-refs.sh" work.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
* master: (72 commits)
runstatus: do not recurse into subdirectories if not needed
grep: fix --fixed-strings combined with expression.
grep: free expressions and patterns when done.
Corrected copy-and-paste thinko in ignore executable bit test case.
An illustration of rev-list --parents --pretty=raw
Allow git-checkout when on a non-existant branch.
gitweb: Decode long title for link tooltips
git-svn: Fix fetch --no-ignore-externals with GIT_SVN_NO_LIB=1
Ignore executable bit when adding files if filemode=0.
Remove empty ref directories that prevent creating a ref.
Use const for interpolate arguments
git-archive: update documentation
Deprecate merge-recursive.py
gitweb: fix over-eager application of esc_html().
Allow '(no author)' in git-svn's authors file.
Allow 'svn fetch' on '(no date)' revisions in Subversion.
git-repack: allow git-repack to run in subdirectory
Remove upload-tar and make git-tar-tree a thin wrapper to git-archive
git-tar-tree: Move code for git-archive --format=tar to archive-tar.c
git-tar-tree: Remove duplicate git_config() call
...
gitweb: Add history and blame to git_difftree_body()
Add blame and history to Deleted files.
Add blame and history to Modified or Type changed files.
Add blame and history to Renamed or Copied files.
This allows us to do
blame->commit->blame->commit->blame->...
instead of
blame->commit->file->blame->commit->file->blame->...
which is longer and easier to get wrong.
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Remove excessively redundant entries from git_difftree_body
1) All entries on the left are blobs and clicking on them
leads to blobs. No more diff or blob depending on what
happened (modified or mode changed) to the file -- this goes
to the right, in the "link" column.
2) Remove redundant "blob" from the link column on the right.
This can now be had by clicking on the entry itself.
This reduces and simplifies the code.
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Added interpolation values for the IP address, port and
canonical hostname of the server as it is contacted and
named by the client and passed in via the extended args.
Added --listen=host_or_ipaddr option suport. Renamed port
variable as "listen_port" correspondingly as well.
Documented mutual exclusivity of --inetd option with
--user, --group, --listen and --port options.
Added compat/inet_pton.c from Paul Vixie as needed.
Small memory leaks need to be cleaned up still.
Signed-off-by: Jon Loeliger <jdl@jdl.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
grep: fix --fixed-strings combined with expression.
"git grep --fixed-strings -e GIT --and -e VERSION .gitignore"
misbehaved because we did not notice this needs to grab lines
that have the given two fixed strings at the same time.
and then grafts them together to make B0 a child of A2. So
the final history looks like (time flows from top to bottom):
true parent touches subdir?
A0 none yes (creates it)
A1 A0 no
A2 A1 yes
B0 none yes (different from what's in A2)
B1 B0 no
B2 B1 yes
"git rev-list --parents --pretty=raw B2" would give "fake"
parents on the "commit " header lines while "parent " header
lines show the parent as recorded in the commit object (i.e. B0
appears to have A2 as its parent on "commit " header but there
is no "parent A2" header line in it).
When you have path limiters, we simplify history to omit
commits that do not affect the specified paths.
So "git rev-list --parents --pretty=raw B2 subdir" would return
"B2 B0 A2 A0" (because B1 and A1 do not touch the path). When
it does so, the "commit " header lines have "fake" parents
(i.e. B2 appears to have B0 as its parent on "commit " header),
but you can still get the true parents by looking at "parent "
header.
This converts receive-pack to use the standard ref locking code
instead of its own. As a side effect, it automatically records
the "push" event to ref-log if enabled.
This updates the use of git-update-ref in git-branch, git-tag
and git-commit to make them safer in a few corner cases as
demonstration.
- git-tag makes sure that the named tag does not exist, allows
you to edit tag message and then creates the tag. If a tag
with the same name was created by somebody else in the
meantime, it used to happily overwrote it. Now it notices
the situation.
- git-branch -d and git-commit (for the initial commit) had the
same issue but with smaller race window, which is plugged
with this.
This adds -d flag to update-ref to allow safe deletion of ref.
Before deleting it, the command checks if the given <oldvalue>
still matches the value the caller thought the ref contained.
Similarly, it also accepts 0{40} or an empty string as <oldvalue>
to allow safe creation of a new ref.
Merge branch 'jc/lt-ref2-with-lt-refs' into jc/ref-locking
* jc/lt-ref2-with-lt-refs:
Fix show-ref usage for --dereference.
Document git-show-ref [-s|--hash] option.
Add man page for git-show-ref
gitignore: git-show-ref is a generated file.
Use Linus' show ref in "git-branch.sh".
Add [-s|--hash] option to Linus' show-ref.
Teach "git checkout" to use git-show-ref
Add "git show-ref" builtin command
I've seen some users get into situtations where their HEAD
symbolic-ref is pointing at a non-existant ref. (Sometimes this
happens during clone when the remote repository lacks a 'master'
branch.) If this happens the user is unable to use git-checkout
to switch branches as there is no prior commit to merge from.
So instead of giving the user low-level errors about how HEAD
can't be resolved and how not a single revision was given change
the type of checkout to be a force and go through with the user's
request anyway.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Quote filename in HTTP Content-Disposition: header
Finish work started by commit a2f3db2 (although not documented
in commit message) of quoting using quotemeta the filename in
HTTP -content_disposition header.
Just in case filename contains end of line character.
Also use consistent coding style to compute -content_disposition
parameter.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Add git_url subroutine, and use it to quote full URLs
Add git_url subroutine, which does what git_param did before commit a2f3db2f5de2a3667b0e038aa65e3e097e642e7d, and is used to quote full
URLs, currently only $home_link.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Split validate_input into validate_pathname and validate_refname
Split validate_input subroutine into validate_pathname which is used
for $project, $file_name and $file_parent parameters, and
validate_refname which is used for $hash, $hash_base, $hash_parent and
$hash_parent_base parameters. Reintroduce validation of $file_name
and $file_parent parameters, removed in a2f3db2f
validate_pathname in addition to what validate_input did checks also
for doubled slashes and NUL character. It does not check if input is
textual hash, and does not check if all characters are from the
following set: [a-zA-Z0-9_\x80-\xff\ \t\.\/\-\+\#\~\%].
validate_refname first check if the input is textual hash, then checks
if it is valid pathname, then checks for invalid characters (according
to git-check-ref-format manpage). It does not check if all charactes
are from the [a-zA-Z0-9_\x80-\xff\ \t\.\/\-\+\#\~\%] set.
We do not have to validate pathnames we got from git.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: extend blame to show links to diff and previous
git_blame2() now has two more columns, "Prev" and "Diff",
before the "Commit" column, as follows:
Prev Diff Commit Line Data
SHA Diff SHA N ...
...
The "Prev" column shows the SHA of the parent commit,
between which this line changed. Clicking on it shows the
blame of the file as of the parent commit, for that line.
So clicking repeatedly on "Prev" would show you the blame
of that file, from the point of view of the changes
of that particular line whose "Prev" you're clicking on.
The "Diff" column shows "Diff" which is a link to blobdiff
between "Prev" and "Commit" commits _for that line_.
So clicking on "Diff" would show you the blobdiff (HTML)
between the parent commit and this commit which changed
that particular line.
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
This is a simple one liner to decode long title string in perl's
internal form to utf-8 for link tooltips.
This is not crucial if the commit message is all in ASCII, however, if
you decide to use other encoding, such as UTF-8, tooltips ain't
readable any more.
Signed-off-by: Yasushi SHOJI <yashi@atmark-techno.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Ignore executable bit when adding files if filemode=0.
If the user has configured core.filemode=0 then we shouldn't set
the execute bit in the index when adding a new file as the user
has indicated that the local filesystem can't be trusted.
This means that when adding files that should be marked executable
in a repository with core.filemode=0 the user must perform a
'git update-index --chmod=+x' on the file before committing the
addition.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
The "list" table element of tree view is identical
to the "blob" link part of the link table element.
I.e. clicking on "blob" is identical to clicking on
the entry itself.
Thus, eliminate "blob" from being shown -- the user
can get identical result by simply clicking on the
entry itself.
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>