t9129-git-svn-i18n-commitencoding: Make compare_svn_head_with() compatible with OSX sed
The sed call used in compare_svn_head_with() uses the + quantifier, which
is not supported in the OSX version of sed. It is replaced by the
equivalent \{1,\}.
Signed-off-by: Marcel Koeppen <git-dev@marzelpan.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Merge branch 'for-junio' of git://source.winehq.org/~julliard/git/git
* 'for-junio' of git://source.winehq.org/~julliard/git/git:
git.el: Allow to commit even if there are no marked files.
git.el: Add possibility to mark files directly in git-update-status-files.
git.el: Add an insert file command.
git.el: Never clear the status buffer, only update the files.
git.el: Fix git-amend-commit to support amending an initial commit.
git.el: Properly handle merge commits in git-amend-commit.
git.el: Simplify handling of merge heads in the commit log-edit buffer.
git.el: Remove the env parameter in git-call-process and git-call-process-string.
git.el: Improve error handling for commits.
This avoids the following misleading error message:
error: unable to create temporary sha1 filename ./objects/15: File exists
mkstemp can fail for many reasons, one of which, ENOENT, can occur if
the directory for the temp file doesn't exist. create_tmpfile tried to
handle this case by always trying to mkdir the directory, even if it
already existed. This caused errno to be clobbered, so one cannot tell
why mkstemp really failed, and it truncated the buffer to just the
directory name, resulting in the strange error message shown above.
Note that in both occasions that I've seen this failure, it has not been
due to a missing directory, or bad permissions, but some other, unknown
mkstemp failure mode that did not occur when I ran git again. This code
could perhaps be made more robust by retrying mkstemp, in case it was a
transient failure.
Signed-off-by: Joey Hess <joey@kitenet.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git.el: Simplify handling of merge heads in the commit log-edit buffer.
Use a single Merge: header instead of one Parent: header for each
parent, and don't list the current HEAD as a merged head. Support
symbolic references too.
Documentation: tutorial: add information about "git help" at the beginning
Talking about "git help" is useful because it has a few more
features (like when using it without arguments or with "-a") and
it may work on non unix like platforms.
Also add a few links to git-help(1) in "See also" sections.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Documentation: user-manual: add information about "git help" at the beginning
Talking about "git help" is useful because it has a few more
features (like when using it without arguments or with "-a") and
it may work on non unix like platforms.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
builtin-remote.c: plug a small memory leak in get_one_remote_for_updates()
We know that the string pointed at by remote->name won't change. It can
be borrowed as the key in the string_list without copying. Other parts of
existing code such as get_one_entry() already rely on this fact.
* git://repo.or.cz/git-gui:
git-gui: Fix the search bar destruction handler.
Update the po template
git-gui: Implement automatic rescan after Tool execution.
git-gui: Allow Tools request arguments from the user.
git-gui: Add a Tools menu for arbitrary commands.
git-gui: Fix the after callback execution in rescan.
git-gui: Implement system-wide configuration handling.
git-gui: try to provide a window icon under X
Since delete_this is an ordinary function, it
should not be passed to cb; otherwise it produces
errors when blame windows are closed. Unfortunately,
it is not noticeable when blame is shown in the
master window, so I missed this bug.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
git-gui: Implement automatic rescan after Tool execution.
The Tools menu is generally intended for commands that
affect the working directory or repository state. Thus,
the user would usually want to initiate rescan after
execution of a tool. This commit implements it.
In case somebody would want to avoid rescanning after
certain tools, it also adds an option that controls it,
although it is not made available through the Add dialog.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
git-gui: Allow Tools request arguments from the user.
While static commands are already useful, some tools need
additional parameters to reach maximum usability. This
commit adds support for passing them one revision name
parameter, and one arbitrary string. With this addition,
the tools menu becomes flexible enough to implement basic
rebase support:
[core]
editor = kwrite
[guitool "Rebase/Abort"]
cmd = git rebase --abort
confirm = yes
[guitool "Rebase/Continue"]
cmd = git rebase --continue
[guitool "Rebase/Skip Commit"]
cmd = git rebase --skip
confirm = yes
[guitool "Rebase/Start..."]
cmd = git rebase $ARGS $REVISION $CUR_BRANCH
title = Start Rebase
prompt = Rebase Current Branch
argprompt = Flags
revprompt = New Base
revunmerged = yes
Some of the options, like title or prompt, are intentionally
not included in the Add dialog to avoid clutter. Also, the
dialog handles argprompt and revprompt as boolean vars.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Due to the emphasis on scriptability in the git
design, it is impossible to provide 100% complete
GUI. Currently unaccounted areas include git-svn
and other source control system interfaces, TopGit,
all custom scripts.
This problem can be mitigated by providing basic
customization capabilities in Git Gui. This commit
adds a new Tools menu, which can be configured
to contain items invoking arbitrary shell commands.
The interface is powerful enough to allow calling
both batch text programs like git-svn, and GUI editors.
To support the latter use, the commands have access
to the name of the currently selected file through
the environment.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
git-gui: Fix the after callback execution in rescan.
The rescan function receives a callback command
as its parameter, which is supposed to be executed
after the scan finishes. It is generally used to
update status. However, rescan may initiate a
loading of a diff, which always calls ui_ready after
completion. If the after handler is called before
that, ui_ready will override the new status.
This commit ensures that the after callback is
properly threaded through the diff machinery.
Since it uncovered the fact that force_first_diff
actually didn't work due to an undeclared global
variable, and the desired effects appeared only
because of the race condition between the diff
system and the rescan callback, I also reimplement
this function to make it behave as originally
intended.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
git-gui: Implement system-wide configuration handling.
With the old implementation any system-wide options appear
to be set locally in the current repository. This commit
adds explicit handling of system options, essentially
interpreting them as customized default_config.
The difficulty in interpreting system options stems from
the fact that simple 'git config' lists all values, while
'git config --global' only values set in ~/.gitconfig,
excluding both local and system options.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
When running under X, we try to set up a window icon by providing a
hand-crafted 16x16 Tk photo image equivalent to the .ico. Wrap in a
catch because the earlier Tcl/Tk 8.4 releases didn't provide the 'wm
iconphoto' command.
Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com> Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The space between the commit and the source attribute is not easily
machine-parseable: if we combine --source with --parents and give a SHA1
as a starting point, it's unnecessarily hard to see where the list of
parents ends and the source decoration begins.
Example:
git show --parents --source $(git rev-list HEAD)
which is admittedly contrived, but can easily happen in scripting.
So use a <tab> instead of a space as the source separator.
The other decorations didn't have this issue, because they were surrounded
by parenthesis, so it's obvious that they aren't parent SHA1's.
It so happens that _visually_ this makes no difference for "git log
--source", since "commit <40-char SHA1>" is 47 characters, so both a space
and a <tab> will end up showing as a single commit. Of course, with
'--pretty=oneline' or '--parents' or '--abbrev-commit' you'll see the
difference.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach ls-files --with-tree=<tree> to work with options other than -c
Originally --with-tree=<tree> was designed for the sole purpose of
checking if a given pathspec makes sense as a parameter to git-commit
using it in conjunction with --error-unmatch. It had logic to avoid
showing the same entry (one came from the original index, another from the
overlayed tree) twice so that it works with -c (i.e. "show-cached"), but
otherwise it was not designed to work with the flags such as -m, -d, etc.
This teaches the same logic to cover the codepath for -m and -d.
* maint:
Documentation: git-svn: fix example for centralized SVN clone
Documentation: fix links to "everyday.html"
revision.c: use proper data type in call to sizeof() within xrealloc
Documentation: git-svn: fix example for centralized SVN clone
The example that tells users how to centralize the effort of the initial
git svn clone operation doesn't work properly. It uses rebase but that
only works if HEAD exists. This adds one extra command to create a
somewhat sensible HEAD that should work in all cases.
Signed-off-by: Jan Krüger <jk@jk.gs> Acked-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
repack: only unpack-unreachable if we are deleting redundant packs
The -A option calls pack-objects with the --unpack-unreachable option so
that the unreachable objects in local packs are left in the local object
store loose. But if the -d option to repack was _not_ used, then these
unpacked loose objects are redundant and unnecessary.
Update tests in t7701.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* git://git.bogomips.org/git-svn:
git-svn: Update git-svn to use the ability to place temporary files within repository directory
Git.pm: Make _temp_cache use the repository directory
git-svn: proper detection of bare repositories
git-svn: respect i18n.commitencoding config
git-svn: don't escape tilde ('~') for http(s) URLs
Git.pm: Make _temp_cache use the repository directory
Update the usage of File::Temp->tempfile to place the temporary files
within the repository directory instead of just letting Perl decide what
directory to use, given there is a repository specified when requesting
the temporary file.
This is needed to be able to fix git-svn on msys as msysperl generates
paths with UNIX-style paths (/tmp/xxx) while the git tools expect natvie
path format (c:/..). The repository dir is stored in native format so by
using it as the base directory for temporary files we always get a
usable native full path.
Signed-off-by: Marten Svanfeldt <developer@svanfeldt.com> Acked-by: Eric Wong <normalperson@yhbt.net>
When in a bare repository (or .git, for that matter), git-svn would fail
to initialise properly, since git rev-parse --show-cdup would not output
anything. However, git rev-parse --show-cdup actually returns an error
code if it's really not in a git directory.
Fix the issue by checking for an explicit error from git rev-parse, and
setting $git_dir appropriately if instead it just does not output.
Signed-off-by: Deskin Miller <deskinm@umich.edu> Acked-by: Eric Wong <normalperson@yhbt.net>
SVN itself always stores log messages in the repository as
UTF-8. git always stores/retrieves everything as raw binary
data with no transformations whatsoever.
To interact with SVN, we need to encode log messages as UTF-8
before sending them to SVN, as SVN cannot do it for us. When
retrieving log messages from SVN, we also need to (attempt to)
reencode the UTF-8 log message back to the user-specified commit
encoding.
Note, handling i18n.logoutputencoding for "git svn log" also
needs to be done in a future change.
Also, this change only deals with the encoding of commit
messages and nothing else (path names, blob content, ...).
In-Reply-To: <8b168cfb0810282014r789ac01dnec51824de1078f0@mail.gmail.com>
James North <tocapicha@gmail.com> wrote:
> Hi,
>
> I'm using git-svn on a system with ISO-8859-1 encoding. The problem is
> when I try to use "git svn dcommit" to send changes to a remote svn
> (also ISO-8859-1).
>
> Seems like git-svn is sending commit messages with utf-8 (just a
> guessing...) and they look bad on the remote svn log. E.g. "Ca?\241a
> de cami?\243n"
>
> I have tried using i18n.commitencoding=ISO-8859-1 as suggested by the
> warning when doing "git svn dcommit" but messages still are sent with
> wrong encoding.
git-svn: don't escape tilde ('~') for http(s) URLs
Thanks to Jose Carlos Garcia Sogo and Björn Steinbrink for the
bug report.
On 2008.10.18 23:39:19 +0200, Björn Steinbrink wrote:
> Hi,
>
> Jose Carlos Garcia Sogo reported on #git that a git-svn clone of this
> svn repo fails for him:
> https://sucs.org/~welshbyte/svn/backuptool/trunk
>
> I can reproduce that here with:
> git-svn version 1.6.0.2.541.g46dc1.dirty (svn 1.5.1)
>
> The error message I get is:
> Apache got a malformed URI: Unusable URI: it does not refer to this
> repository at /usr/local/libexec/git-core/git-svn line 4057
>
> strace revealed that git-svn url-encodes ~ while svn does not do that.
>
> For svn we have:
> write(5, "<S:update-report send-all=\"true\" xmlns:S=\"svn:\">
> <S:src-path>https://sucs.org/~welshbyte/svn/backuptool/trunk</S:src-path>...
>
> While git-svn shows:
> write(7, "<S:update-report send-all=\"true\" xmlns:S=\"svn:\">
> <S:src-path>https://sucs.org/%7Ewelshbyte/svn/backuptool/trunk</S:src-path>...
date/time: do not get confused by fractional seconds
The date/time parsing code was confused if the input time HH:MM:SS is
followed by fractional seconds. Since we do not record anything finer
grained than seconds, we could just drop fractional part, but there is a
twist.
We have taught people that not just spaces but dot can be used as word
separators when spelling things like:
$ git log --since 2.days
$ git show @{12:34:56.7.days.ago}
and we shouldn't mistake "7" in the latter example as a fraction and
discard it.
The rules are:
- valid days of month/mday are always single or double digits.
- valid years are either two or four digits
No, we don't support the year 600 _anyway_, since our encoding is based
on the UNIX epoch, and the day we worry about the year 10,000 is far
away and we can raise the limit to five digits when we get closer.
- Other numbers (eg "600 days ago") can have any number of digits, but
they cannot start with a zero. Again, the only exception is for
two-digit numbers, since that is fairly common for dates ("Dec 01" is
not unheard of)
So that means that any milli- or micro-second would be thrown out just
because the number of digits shows that it cannot be an interesting date.
A milli- or micro-second can obviously be a perfectly fine number
according to the rules above, as long as it doesn't start with a '0'. So
if we have
12:34:56.123
then that '123' gets parsed as a number, and we remember it. But because
it's bigger than 31, we'll never use it as such _unless_ there is
something after it to trigger that use.
So you can say "12:34:56.123.days.ago", and because of the "days", that
123 will actually be meaninful now.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* git://git.kernel.org/pub/scm/gitk/gitk:
gitk: Fix linehtag undefined error with file highlighting
gitk: Fix commit encoding support
gitk: Fix transient windows on Win32 and MacOS
gitk: Add accelerators to frequently used menu commands
gitk: Implement a user-friendly Edit View dialog
gitk: Improve cherry-pick error handling
gitk: Make cherry-pick call git-citool on conflicts
gitk: Make gitk dialog windows transient
gitk: Add Return and Escape bindings to dialogs
gitk: Cope with unmerged files in local changes
gitk: Make "show origin of this line" work on fake commits
gitk: Unify handling of merge diffs with normal 2-way diffs
gitk: Make the background color of marked lines configurable
gitk: Add a menu item to show where a given line comes from
gitk: Fix some off-by-one errors in computing which line to blame
gitk: Allow starting gui blame for a specific line
gitk: Fix file list context menu for merge commits
gitk: Allow forcing branch creation if it already exists
gitk: Fix linehtag undefined error with file highlighting
Occasionally gitk will throw a Tcl error complaining that linehtag(n)
is undefined when. It happens when the commit list is still growing
(e.g. when updating the commit list) and gitk is set to highlight
commits that affect certain file(s). What happens is that the changes
to the commit list set need_redisplay to indicate that the display
needs to be redrawn. That causes the next call to drawcommits to call
clear_display, which unsets iddrawn and thus ensures that readfhighlight
won't call bolden on any rows that have moved. However, it is possible
for readfhighlight to be called after the commit list has changed but
before drawcommits has run, meaning that readfhighlight will potentially
think that rows have been drawn when they haven't, because of the
change in the id -> row mapping (and the fact that iddrawn is indexed
by id but line[hnd]tag are indexed by row number).
This fixes it (and also optimizes things a little) by making bolden
and bolden_name check need_redisplay before doing anything. If
need_redisplay is set, then there is no point doing anything because
the whole display is about to get cleared and redrawn, and it avoids
looking up line[hn]tag using stale row numbers.
This commit fixes two problems with commit encodings:
1) git-log actually uses i18n.logoutputencoding to generate
its output, and falls back to i18n.commitencoding only
when that option is not set. Thus, gitk should use its
value to read the results, if available.
2) The readcommit function did not process encodings at all.
This led to randomly appearing misconverted commits if
the commit encoding differed from the current locale.
Now commit messages should be displayed correctly, except
when logoutputencoding is set to an encoding that cannot
represent charecters in the message. For example, it is
impossible to convert Japanese characters from Shift-JIS
to CP-1251 (although the reverse conversion works).
The reason for using git log to read the commit and then getting
Tcl to convert its output is that is essentially what happens in
the normal path through getcommitlines, hence there is less chance
for unintended differences in how commits are processed in
getcommitlines and do_readcommit.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Paul Mackerras <paulus@samba.org>
Transient windows cause problems on these platforms:
- On Win32 the windows appear in the top left corner
of the screen. In order to fix it, this patch causes
them to be explicitly centered on their parents by
an idle handler.
- On MacOS with Tk 8.4 they appear without a title bar.
Since it is clearly unacceptable, this patch disables
transient on that platform.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Paul Mackerras <paulus@samba.org>
gitk: Add accelerators to frequently used menu commands
This commit documents keyboard accelerators used for menu
commands in the menu, as it is usually done, and adds some
more, e.g. F4 to invoke Edit View (or New View if the current
view is the un-editable "All files" view).
The changes include a workaround for handling Shift-F4 on
systems where XKB binds special XF86_Switch_VT_* symbols
to Ctrl-Alt-F* combinations. Tk often receives these codes
when Shift-F* is pressed, so it is necessary to bind the
relevant actions to them as well.
Signed-off-by: Alexander Gavrilov <angavrilov@gmail.com> Signed-off-by: Paul Mackerras <paulus@samba.org>
* gb/gitweb-snapshot-pathinfo:
gitweb: embed snapshot format parameter in PATH_INFO
gitweb: retrieve snapshot format from PATH_INFO
gitweb: make the supported snapshot formats array global
* 'ds/uintmax-config' (early part):
Add autoconf tests for pthreads
Make Pthread link flags configurable
Add Makefile check for FreeBSD 4.9-SECURITY
Build: add NO_UINTMAX_T to support ancient systems
* np/pack-safer:
t5303: fix printf format string for portability
t5303: work around printf breakage in dash
pack-objects: don't leak pack window reference when splitting packs
extend test coverage for latest pack corruption resilience improvements
pack-objects: allow "fixing" a corrupted pack without a full repack
make find_pack_revindex() aware of the nasty world
make check_object() resilient to pack corruptions
make packed_object_info() resilient to pack corruptions
make unpack_object_header() non fatal
better validation on delta base object offsets
close another possibility for propagating pack corruption
* bc/maint-keep-pack:
t7700: test that 'repack -a' packs alternate packed objects
pack-objects: extend --local to mean ignore non-local loose objects too
sha1_file.c: split has_loose_object() into local and non-local counterparts
t7700: demonstrate mishandling of loose objects in an alternate ODB
builtin-gc.c: use new pack_keep bitfield to detect .keep file existence
repack: do not fall back to incremental repacking with [-a|-A]
repack: don't repack local objects in packs with .keep file
pack-objects: new option --honor-pack-keep
packed_git: convert pack_local flag into a bitfield and add pack_keep
t7700: demonstrate mishandling of objects in packs with a .keep file
* mv/remote-rename:
git-remote: document the migration feature of the rename subcommand
git-remote rename: migrate from remotes/ and branches/
remote: add a new 'origin' variable to the struct
Implement git remote rename
* lt/decorate:
rev-list documentation: clarify the two parts of history simplification
Document "git log --simplify-by-decoration"
Document "git log --source"
revision traversal: '--simplify-by-decoration'
Make '--decorate' set an explicit 'show_decorations' flag
revision: make tree comparison functions take commits rather than trees
Add a 'source' decorator for commits
* git://repo.or.cz/git-gui:
git-gui: Request blame metadata in utf-8.
git-gui: Add the Show SSH Key item to the clone dialog.
git-gui: Fix focus transition in the blame viewer.
t7700: test that 'repack -a' packs alternate packed objects
Previously, when 'repack -a' was called and there were no packs in the local
repository without a .keep file, the repack would fall back to calling
pack-objects with '--unpacked --incremental'. This resulted in the created
pack file, if any, to be missing the packed objects in the alternate object
store. Test that this specific case has been fixed.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil> Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint:
Start 1.6.0.5 cycle
Fix pack.packSizeLimit and --max-pack-size handling
checkout: Fix "initial checkout" detection
Remove the period after the git-check-attr summary
Fix pack.packSizeLimit and --max-pack-size handling
If the limit was sufficiently low, having a single object written
could bust the limit (by design), but caused the remaining allowed
size to go negative for subsequent objects, which for an unsigned
variable is a rather huge limit.
Signed-off-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
rev-list documentation: clarify the two parts of history simplification
One set of options and parameters determine what commits are involved in
the simplification process, and another set of options determine how the
simplification is done. Clarify their distinction at the beginning.
Signed-off-by: Santi Béjar <santi@agolina.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Earlier commit 5521883 (checkout: do not lose staged removal, 2008-09-07)
tightened the rule to prevent switching branches from losing local
changes, so that staged removal of paths can be protected, while
attempting to keep a loophole to still allow a special case of switching
out of an un-checked-out state.
However, the loophole was made a bit too tight, and did not allow
switching from one branch (in an un-checked-out state) to check out
another branch.
The change to builtin-checkout.c in this commit loosens it to allow this,
by not insisting the original commit and the new commit to be the same.
It also introduces a new function, is_index_unborn (and an associated
macro, is_cache_unborn), to check if the repository is truly in an
un-checked-out state more reliably, by making sure that $GIT_INDEX_FILE
did not exist when populating the in-core index structure. A few places
the earlier commit 5521883 added the check for the initial checkout
condition are updated to use this function.
When the "-v" option is given, we put diff of what is to be committed into
the commit template, and then strip it back out again after the user has
edited it.
We used to look for the diff by searching for the "diff --git a/"
header. With diff.mnemonicprefix set in the configuration, however, this
pattern does not match. The pattern is loosened to cover this case.
Also, if the user puts their own diff in the message (e.g., as a sample
output), then we will accidentally trigger the pattern, removing part of
their output.
We can avoid doing this stripping altogether if the user didn't use "-v"
in the first place, so we know that any match we find will be a false
positive.
[jc: this fix was split out of a series originally meant for master.]
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we showed the initial commit, we had no reference to
diff against, so we went through the cache manually.
Nowadays, however, we have a virtual empty tree commit, so
we can simply diff against that to get the same results.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
repack: do not fall back to incremental repacking with [-a|-A]
When repack is called with either the -a or -A option, the user has
requested to repack all objects including those referenced by the
alternates mechanism. Currently, if there are no local packs without
.keep files, then repack will call pack-objects with the
'--unpacked --incremental' options which causes it to exclude alternate
packed objects. So, remove this fallback.
Signed-off-by: Brandon Casey <casey@nrlssc.navy.mil> Signed-off-by: Junio C Hamano <gitster@pobox.com>