[PATCH] If NO_MMAP is defined, fake mmap() and munmap()
Since some platforms do not support mmap() at all, and others do only just
so, this patch introduces the option to fake mmap() and munmap() by
malloc()ing and read()ing explicitely.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
This makes it possible to have a "sparse" git object subdirectory
structure, something that has become much more attractive now that people
use pack-files all the time.
As a result of pack-files, a git object directory doesn't necessarily have
any individual objects lying around, and in that case it's just wasting
space to keep the empty first-level object directories around: on many
filesystems the 256 empty directories will be aboue 1MB of diskspace.
Even more importantly, after you re-pack a project that _used_ to be
unpacked, you could be left with huge directories that no longer contain
anything, but that waste space and take time to look through.
With this change, "git prune-packed" can just do an rmdir() on the
directories, and they'll get removed if empty, and re-created on demand.
This patch also tries to fix up "write_sha1_from_fd()" to use the new
common infrastructure for creating the object files, closing a hole where
we might otherwise leave half-written objects in the object database.
[jc: I unoptimized the part that really removes the fan-out directories
to ease transition. init-db still wastes 1MB of diskspace to hold 256
empty fan-outs, and prune-packed rmdir()'s the grown but empty directories,
but runs mkdir() immediately after that -- reducing the saving from 150KB
to 146KB. These parts will be re-introduced when everybody has the
on-demand capability.]
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-status truncates filenames up to the first occurrence of a whitespace
character when displaying. More precisely, it displays the filename up to any
field seperator defined in $IFS.
This patch fixes a bug in git-ls-tree in which the wrong filenames are
listed if the exact same file and directory contents are present in
another location in the tree.
Added a new series of test cases for directory and filename handling.
Signed-off-by: Robert Fitzsimons <robfitz@273k.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
It reorganizes the code and also has saner command line options
syntax. Unlike git-applymbox, it can take more than one mailbox
file from the command line, as well as reading from the standard
input when '-' is specified.
Add an new option --show-index-info to git-apply command to
summarize the index information new git-diff outputs. The
command shows something similar to git-ls-files --stage output
for the pre-change image:
Show original and resulting blob object info in diff output.
This adds more cruft to diff --git header to record the blob SHA1 and
the mode the patch/diff is intended to be applied against, to help the
receiving end fall back on a three-way merge. The new header looks
like this:
diff --git a/apply.c b/apply.c
index 7be5041..8366082 100644
--- a/apply.c
+++ b/apply.c
@@ -14,6 +14,7 @@
// files that are being modified, but doesn't apply the patch
// --stat does just a diffstat, and doesn't actually apply
+// --show-index-info shows the old and new index info for...
...
Upon receiving such a patch, if the patch did not apply cleanly to the
target tree, the recipient can try to find the matching old objects in
her object database and create a temporary tree, apply the patch to
that temporary tree, and attempt a 3-way merge between the patched
temporary tree and the target tree using the original temporary tree
as the common ancestor.
The patch lifts the code to compute the hash for an on-filesystem
object from update-index.c and makes it available to the diff output
routine.
When you are applying 200 mails in sequence, .dotest/ directory
will be littered with many messsages, and when the patch in one
of them fails to apply, it is not obvious which message was
being processed. Remove the one that has been already dealt
with, so that the last failed one is found typically as the
lowest numbered split message.
Fall back to three-way merge when applying a patch.
After git-apply fails, attempt to find a base tree that the patch
cleanly applies to, and do a three-way merge using that base tree into
the current index, if .dotest/.3way file exists. This flag can be
controlled by giving -m flag to git-applymbox command.
When the fall-back merge fails, the working tree can be resolved the
same way as you would normally hand resolve a conflicting merge.
When making commit, use .dotest/final-commit as the log message
template. Or you could just choose to 'git-checkout-index -f -a'
to revert the failed merge.
The current "git tag -s" thing always uses the tagger name as the signing
user key, which is very irritating, since my key is under my email
address, but the tagger key obviously contains the actual machine name
too.
Now, I could just use "GIT_COMMITTER_EMAIL" and force it to be my real
email, but I actually think that it's nice to see which machine I use for
my work.
So rather than force my tagger ID to have to match the gpg key name, just
support the "-u" flag to "git tag" instead. It implicitly enables signing,
since it doesn't make any sense without it. Thus:
git tag -u <gpg-key-name> <tag-name> [<tagged-object>]
will use the named gpg key for signing.
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
In addition to hardcoded list of kernel people, read from .mailmap file
the list of email-to-name translations. Modernize regexps here and there
minimally while at it.
upload-pack: Do not choke on too many heads request.
Cloning from a repository with more than 256 refs (heads and tags
included) will choke, because upload-pack has a built-in limit of
feeding not more than MAX_NEEDS (currently 256) heads to underlying
git-rev-list. This is a problem when cloning a repository with many
tags, like http://www.linux-mips.org/pub/scm/linux.git, which has 290+
tags.
This commit introduces a new flag, --all, to git-rev-list, to include
all refs in the repository. Updated upload-pack detects requests that
ask more than MAX_NEEDS refs, and sends everything back instead.
We may probably want to tweak the definitions of MAX_NEEDS and
MAX_HAS, but that is a separate topic.
When we updated the marker for new files from 'N' to 'A', we forgot to
notice that the letter is already taken by the All-Or-None mark.
Change the All-Or-None marker to '*' to resolve this conflict.
git-diff-tree -r --diff-filter='R*' -M
shows all the changes (not just renames) that are contained in commits
that have renames, in comparison with:
git-diff-tree -r --diff-filter='R' -M
shows the same set of changes but the diff output are limited only to
renaming changes.
The original plan was to do 3-way merge between local working tree,
index and the patch being applied, but that was never implemented.
Retire the flag to control its behaviour.
git-apply: allow operating in sparsely populated working tree.
This patch teaches 'git-apply --index' to automatically check
out a file being patched. This happens only when the working
tree does not have it checked out.
Return error when not checking out an entry due to dirtiness.
Without -f flag, 'git-checkout-index foo.c' issued an error message
when foo.c already existed in the working tree and did not match index.
However it did not return an error from the underlying checkout_entry()
function and resulted in a successful exit(0).
There was a leftover use of sed that attempted to remove the commit ID
output from git-diff-tree, which turned into an expensive no-op when
git-diff-tree output header format changed about three months ago.
Drop it.
On Cygwin, use symbolic ref, not a symbolic link, to express .git/HEAD
H. Peter Anvin says that Samba "promotes" symlinks to hardlinks while
Cygwin itself uses .lnk files to emulate symlinks. Avoid using symbolic
link for .git/HEAD on Cygwin.
This does not help the symlinks recorded in trees as user data, but
at least we do not use them for our own bookkeeping.
Error message from get_sha1() on ambiguous short SHA1.
Unlike cases where "no such object exists", the case where specified
prefix is ambiguous would confuse the user if we say "no such commit"
or such. Give an extra error message from the uniqueness check if
there are more than one objects that match the given prefix.
[PATCH] Enable and fix support for base less merges.
Let the merge strategies handle the base less case if they are able to
do it. It also fixes git-resolve.sh to die if no common ancestors
exists, instead of doing the wrong thing. Furthermore, it contains a
small independent fix for git-merge.sh and a fix for a base less code
path in gitMergeCommon.py.
With this it's possible to use
git merge -s recursive 'merge message' A B
to do a base less merge of A and B.
[jc: Thanks Fredrik for fixing the brown-paper-bag in git-merge.
I fixed a small typo in git-merge-resolve fix; 'test' equality
check is spelled with single equal sign -- C-style double equal
sign is bashism.]
Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se> Signed-off-by: Junio C Hamano <junkio@cox.net>
Make sure get_sha1 does not accept ambiguous sha1 prefix.
The original code did not even check alternates, and was confused if
an unpacked object was uniquely found when there was another object
that shares the same prefix in the pack.
GIT already did everything I wanted it to do since mid 0.99.7,
and it has almost everything I want it to have now, except a
couple of minor tweaks and enhancements.
The documentation for git-clone is behind the actual command.
I have been getting tired of reading the shell script to see
what the arguments are so here is an update of the actual documentation.
Signed-off-by: Eric Biederman <ebiederman@xmission.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
[PATCH] Teach the recursive merge strategy about renames.
It will now merge cases where a file was renamed in one branch and
modified in the other branch cleanly. We also detect a couple of
conflict cases now that wasn't detected before.
Signed-off-by: Fredrik Kuivinen <freku045@student.liu.se> Signed-off-by: Junio C Hamano <junkio@cox.net>
This makes "git-update-index" avoid the new index file write if it didn't
make any changes to the index.
It still doesn't make things like "git status" be read-only operations in
general, but if the index file doesn't need refreshing, it now will at
least avoid making unnecessary changes.
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Instead of "git status" ignoring (and hiding) potential errors from the
"git-update-index" call, make it exit if it fails, and show the error.
In order to do this, use the "-q" flag (to ignore not-up-to-date files)
and add a new "--unmerged" flag that allows unmerged entries in the index
without any errors.
This also avoids marking the index "changed" if an entry isn't actually
modified, and makes sure that we exit with an understandable error message
if the index is corrupt or unreadable. "read_cache()" no longer returns an
error for the caller to check.
Finally, make die() and usage() exit with recognizable error codes, if we
ever want to check the failure reason in scripts.
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
- The location of openssl development files got customizable.
- The location of iconv development files got customizable.
- Pass $TAR down to t5000 test so that the user can override with
'gmake TAR=gtar'.
- Solaris 'bc' does not seem to grok "define abs()". There is no
reason to use bc there -- expr would do.
This adds the counterpart of git-update-ref that lets you read
and create "symbolic refs". By default it uses a symbolic link
to represent ".git/HEAD -> refs/heads/master", but it can be compiled
to use the textfile symbolic ref.
The places that did 'readlink .git/HEAD' and 'ln -s refs/heads/blah
.git/HEAD' have been converted to use new git-symbolic-ref command, so
that they can deal with either implementation.
[PATCH] Allow reading "symbolic refs" that point to other refs
This extends the ref reading to understand a "symbolic ref": a ref file
that starts with "ref: " and points to another ref file, and thus
introduces the notion of ref aliases.
This is in preparation of allowing HEAD to eventually not be a symlink,
but one of these symbolic refs instead.
[jc: Linus originally required the prefix to be "ref: " five bytes
and nothing else, but I changed it to allow and strip any number of
leading whitespaces to match what update-ref.c does.]
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Teach update-ref about a symbolic ref stored in a textfile.
A symbolic ref is a regular file whose contents is "ref:", followed by
optional leading whitespaces, followed by a GIT_DIR relative pathname,
followed by optional trailing whitespaces (the optional whitespaces
are unconditionally removed, so you cannot have leading nor trailing
whitespaces). This can be used in place of a traditional symbolic
link .git/HEAD that usually points at "refs/heads/master". You can
instead have a regular file .git/HEAD whose contents is
"ref: refs/heads/master".
[jc: currently the code does not enforce the symbolic ref to begin with
refs/, unlike the symbolic link case. It may be worthwhile to require
either case to begin with refs/ and not have any /./ nor /../ in them.]
[jc: The original by Linus fetched and overwrote branch heads with
--all, which felt dangerous and wrong, so I removed it. Also this
version does not use any refs that resulted as --tags for later
merge. ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Add the sanity checks discussed on the list with Nick Hengeveld in
<20050927000931.GA15615@reactrix.com>.
* unlink of previous and rename from temp to previous can fail for
reasons other than benign ones (missing previous and missing temp).
Report these failures when we encounter them, to make diagnosing
problems easier.
* when rewinding the partially written result, make sure to
truncate the file.
Also verify the pack after downloading by calling
verify_packfile().
The archive generated with git-tar-tree had 0755 and 0644 mode bits.
This inconvenienced the extractor with umask 002 by robbing g+w bit
unconditionally. Just write it out with loose permissions bits and
let the umask of the extractor do its job.
A patch that contains no actual diff, and that doesn't change any
meta-data is bad. It shouldn't be a patch at all, and git-apply shouldn't
just accept it.
This caused a corrupted patch to be silently applied as an empty change in
the kernel, because the corruption ended up making the patch look empty.
An example of such a patch is one that contains the patch header, but
where the initial fragment header (the "@@ -nr,.." line) is missing,
causing us to not parse any fragments.
The real "patch" program will also flag such patches as bad, with the
message
patch: **** Only garbage was found in the patch input.
and we should do likewise.
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
After seeing Jeff's guide, I changed my mind about the
big-rename transition plan. Even if Porcelains are kept up to
date, those web documents that describes older world order would
live longer and people will stumble across them via google
searches. And who knows how many mirrored copies there are.
The backward compatible symbolic links *will* be removed before
1.0. But that will not happen in 0.99.8.
This again makes git-pull to use git-merge, so that different merge
strategy can be specified from the command line. Without explicit
strategy parameter, it defaults to git-merge-resolve if only one
remote is pulled, and git-merge-octopus otherwise, to keep the
default behaviour of the command the same as the original.
Also this brings another usability measure: -n flag from the command
line, if given, is passed to git-merge to prevent it from running the
diffstat at the end of the merge.