Work around curl-gnutls not liking to be reinitialized
curl versions 7.16.3 to 7.18.0 included had a regression in which https
requests following curl_global_cleanup/init sequence would fail with ASN1
parser errors with curl-gnutls. Such sequences happen in some cases such
as git fetch.
We work around this by removing the http_init and http_cleanup calls from
get_refs_via_curl, replacing them with a transport->data initialization
with the http_walker (which does http_init).
While the http_walker is not currently used in get_refs_via_curl, http
and walker code refactor will make it use it.
Signed-off-by: Mike Hommey <mh@glandium.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jakub Narebski <jnareb@gmail.com> wrote Sun, Feb 03, 2008:
> Junio C Hamano wrote:
> > Jakub Narebski <jnareb@gmail.com> writes:
> >
> > [From] http://thread.gmane.org/gmane.comp.version-control.git/53457/focus=53458
> Julian Phillips:
> > Are you using docbook xsl 1.72? There are known problems building the
> > manpages with that version. 1.71 works, and 1.73 should work when it get
> > released.
I was able to solve this problem with this patch, which adds a XSL file
used specifically for DOCBOOK_XSL_172=YesPlease and where dots and
backslashes are escaped properly so they won't be substituted to the
wrong thing further down the "DocBook XSL pipeline". Doing the escaping
in the existing callout.xsl breaks v1.70.1. Hopefully v1.73 will end
this part of the manpage nightmare.
builtin-commit: remove .git/SQUASH_MSG upon successful commit
After doing a merge --squash, and commit afterwards, the commit message
template SQUASH_MSG in the git directory is not removed, which means that
the content of SQUASH_MSG is used as default commit message for all
subsequent commits. So have git commit remove the file SQUASH_MSG from
the git directory upon a successful commit.
The problem was discovered by Frédéric Brière, reported through
http://bugs.debian.org/464656
Signed-off-by: Gerrit Pape <pape@smarden.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make git prune remove temporary packs that look like write failures
Write errors when repacking (eg, due to out-of-space conditions)
can leave temporary packs (and possibly other files beginning
with "tmp_") lying around which no existing
codepath removes and which aren't obvious to the casual user.
These can also be multi-megabyte files wasting noticeable space.
Unfortunately there's no way to definitely tell in builtin-prune
that a tmp_ file is not being used by a concurrent process,
such as a fetch. However, it is documented that pruning should
only be done on a quiet repository and --expire is honoured
(using code from Johannes Schindelin, along with a test case
he wrote) so that its safety is the same as that of loose
object pruning.
Since they might be signs of a problem (unlike orphaned loose
objects) the names of any removed files are printed.
Signed-off-by: David Tweed (david.tweed@gmail.com) Acked-by: Nicolas Pitre <nico@cam.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Document that the default of branch.autosetupmerge is true
In 34a3e69 (git-branch: default to --track) the default was changed to
true, to help new git users. But yours truly forgot to update the
documentation. This fixes it.
Noticed by Kalle Olavi Niemitalo.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
find_beginning_of_line didn't take into account that the
previous line might have ended with \ in which case it shouldn't
stop but continue its search.
Signed-off-by: Frank Lichtenheld <frank@lichtenheld.de> Acked-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Currently, the --pretty=format prefix is looked up in a
tight loop in strbuf_expand(), if prefix is found it is then
used as argument for format_commit_item() that does another
search by a switch statement to select the proper operation.
Because the switch statement is already able to discard
unknown matches we don't need the prefix lookup before
to call format_commit_item().
Signed-off-by: Marco Costalba <mcostalba@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Work around curl-gnutls not liking to be reinitialized
curl versions 7.16.3 to 7.18.0 included had a regression in which https
requests following curl_global_cleanup/init sequence would fail with ASN1
parser errors with curl-gnutls. Such sequences happen in some cases such
as git fetch.
We work around this by removing the http_init and http_cleanup calls from
get_refs_via_curl, replacing them with a transport->data initialization
with the http_walker (which does http_init).
While the http_walker is not currently used in get_refs_via_curl, http
and walker code refactor will make it use it.
Signed-off-by: Mike Hommey <mh@glandium.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git pack-objects" has the option --max-pack-size to limit the file
size of the packs to a certain amount of bytes. On platforms where
the pack file size is limited by filesystem constraints, it is easy
to forget this option, and this option does not exist for "git gc"
to begin with.
So introduce a config variable to set the default maximum, but make
this overrideable by the command line.
Suggested by Tor Arvid Lund.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
gitweb: Use the config file to set repository owner's name.
Now gitweb checks if gitweb.owner exists before trying to get filesystem's
owner.
Allow to use configuration variable gitweb.owner set the repository owner,
it checks the gitweb.owner, if not set it uses filesystem directory's owner.
Useful when we don't want to maintain project list file, and all
repository directories have to have the same owner (for example when the
same SSH account is shared for all projects, using ssh_acl to control
access instead).
Signed-off-by: Bruno Ribas <ribas@c3sl.ufpr.br> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-web--browse: do not start the browser with nohup
There is no good reason to run GUI browsers using "nohup". It does not
solve any real problem but creates annoying "nohup.out" files in every
directory where git help -w is run.
Signed-off-by: Dmitry Potapov <dpotapov@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
If the test failed, it was giving really unclear ed script
output. Instead, give a diff that sort of suggests the problem. Also
replaces the use of "git diff" for this purpose with "diff -u".
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
You can also create branches, in exactly the same way, with checkout -b.
This introduces branch.{c,h} library files for doing porcelain-level
operations on branches (such as creating them with their appropriate
default configuration).
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
This makes write_tree_from_memory(), which writes the active cache as
a tree and returns the struct tree for it, available to other code. It
also makes available merge_trees(), which does the internal merge of
two trees with a known base, and merge_recursive(), which does the
recursive internal merge of two commits with a list of common
ancestors.
The first two of these will be used by checkout -m, and the third is
presumably useful in general, although the implementation of checkout
-m which entirely matches the behavior of the shell version does not
use it (since it ignores the difference of ancestry between the old
branch and the new branch).
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
This option allows the caller to reset everything that isn't unmerged,
leaving the unmerged things to be resolved. If, after a merge of
"working" and "HEAD", this is used with "HEAD" (reset, !update), the
result will be that all of the changes from "local" are in the working
tree but not added to the index (either with the index clean but
unchanged, or with the index unmerged, depending on whether there are
conflicts).
This will be used in checkout -m.
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Discard "deleted" cache entries after using them to update the working tree
Way back in read-tree.c, we used a mode 0 cache entry to indicate that
an entry had been deleted, so that the update code would remove the
working tree file, and we would just skip it when writing out the
index file afterward.
These days, unpack_trees is a library function, and it is still
leaving these entries in the active cache. Furthermore, unpack_trees
doesn't correctly ignore those entries, and who knows what other code
wouldn't expect them to be there, but just isn't yet called after a
call to unpack_trees. To avoid having other code trip over these
entries, have check_updates() remove them after it removes the working
tree files.
While we're at it, simplify the loop in check_updates(), and avoid
passing global variables as parameters to check_updates(): there is
only one call site anyway.
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
(This applies only to errors where a plausible operation is impossible due
to the particular data, not to errors resulting from misuse of the merge
functions.)
This will allow builtin-checkout to suppress merge errors if it's
going to try more merging methods.
Additionally, if unpack_trees() returns with an error, but without
printing anything, it will roll back any changes to the index (by
rereading the index, currently). This obviously could be done by the
caller, but chances are that the caller would forget and debugging
this is difficult. Also, future implementations may give unpack_trees() a
more efficient way of undoing its changes than the caller could.
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
Return an error from unpack_trees() instead of calling die(), and exit
with an error in read-tree, builtin-commit, and diff-lib. merge-recursive
already expected an error return from unpack_trees, so it doesn't need to
be changed. The merge function can return negative to abort.
This will be used in builtin-checkout -m.
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org>
In ba227857(Reduce the number of connects when fetching), we checked
the return value of git_connect() to see if the connection was
successful.
However, for the git:// protocol, there is no need to have another
process, so the return value was NULL.
Now, it makes sense to assume the rule that git_connect() will return
NULL if it fails (at the moment, it die()s if it fails), so return
a dummy child process.
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Acked-by: Daniel Barkalow <barkalow@iabervon.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refuse to revert a file if it is modified in an existing buffer but
not saved. On success, revert the buffers that contains the files that
have been reverted.
Signed-off-by: Alexandre Julliard <julliard@winehq.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git.el: Support for showing unknown/ignored directories.
Instead of recursing into directories that only contain unknown files,
display only the directory itself. Its contents can be expanded with
git-find-file (bound to C-m).
Signed-off-by: Alexandre Julliard <julliard@winehq.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
There was an embarrassing pair of off-by-one miscounting that
failed to match path "a/b/c" when "a/.gitattributes" tried to
name it with relative path "b/c".
Improve bash prompt to detect various states like an unfinished merge
This patch makes the git prompt (when enabled) show if a merge or a
rebase is unfinished. It also detects if a bisect is being done as
well as detached checkouts.
An uncompleted git-am cannot be distinguised from a rebase (the
non-interactive version). Instead of having an even longer prompt
we simply ignore that and hope the power users that use git-am knows
the difference.
Signed-off-by: Robin Rosenberg <robin.rosenberg@dewire.com> Acked-by: Shawn O. Pearce <spearce@spearce.org>
compat: Add simplified merge sort implementation from glibc
qsort in Windows 2000 (and various other C libraries) is a Quicksort
with the usual O(n^2) worst case. Unfortunately, sorting Git trees
seems to get very close to that worst case quite often:
This patch adds a simplified version of the merge sort that is glibc's
qsort(3). As a merge sort, this needs a temporary array equal in size
to the array that is to be sorted, but has a worst-case performance of
O(n log n).
The complexity that was removed is:
* Doing direct stores for word-size and -aligned data.
* Falling back to quicksort if the allocation required to perform the
merge sort would likely push the machine into swap.
Even with these simplifications, this seems to outperform the Windows
qsort(3) implementation, even in Windows XP (where it is "fixed" and
doesn't trigger O(n^2) complexity on trees).
[jes: moved into compat/qsort.c, as per Johannes Sixt's suggestion]
[bcd: removed gcc-ism, thanks to Edgar Toernig. renamed make variable
per Junio's comment.]
Signed-off-by: Brian Downing <bdowning@lavos.net> Signed-off-by: Steffen Prohaska <prohaska@zib.de> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously, we set the GIT_CONFIG environment variable in
our tests so that only that file was read. However, setting
it to a static value is not correct, since we are not
necessarily always in the same directory; instead, we want
the usual git config file lookup to happen.
To do this, we stop setting GIT_CONFIG, which means that we
must now suppress the reading of the system-wide and user
configs.
This exposes an incorrect test in t1500, which is also
fixed (the incorrect test worked because we were failing to
read the core.bare value from the config file, since the
GIT_CONFIG variable was pointing us to the wrong file).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The GIT_CONFIG_NOGLOBAL and GIT_CONFIG_NOSYSTEM environment
variables are magic undocumented switches that can be used
to ensure a totally clean environment. This is necessary for
running reliable tests, since those config files may contain
settings that change the outcome of tests.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
gitweb: Make feed entries point to commitdiff view
Change feeds entries (feeds items) from pointing (linking) to 'commit'
view to pointing to 'commitdiff' view.
First, feed entries have whatchanged-like list of files which were
modified in a commit, so 'commitdiff' view more naturally reflects
feed entry (is more naturally alternate / extended version of a feed
item). Second, this way the patches are shown directly and code review
is done more easily via watching feeds.
[jn: Rewritten commit message]
Signed-off-by: Florian La Roche <laroche@redhat.com> Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-svn: improve repository URL matching when following parents
This way we can avoid the spawning of a new SVN::Ra session by
reusing the existing one.
The most problematic issue is that some svn servers disallow
too many connections from a single IP, so this will allow
git-svn to fetch from those repositories with a higher success
rate by using fewer connections.
This sometimes showed up as a new (and redundant)
[svn-remote "$parent_refname"] entry in $GIT_DIR/svn/.metadata.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
I was looking at some of the perl commands, and noticed that
git-remote was the only one to lack a 'use strict' pragma at the top,
which could be a good thing for its maintainability. Hopefully, the
required changes are minimal.
Signed-off-by: Rafael Garcia-Suarez <rgarciasuarez@gmail.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
safecrlf: Add mechanism to warn about irreversible crlf conversions
CRLF conversion bears a slight chance of corrupting data.
autocrlf=true will convert CRLF to LF during commit and LF to
CRLF during checkout. A file that contains a mixture of LF and
CRLF before the commit cannot be recreated by git. For text
files this is the right thing to do: it corrects line endings
such that we have only LF line endings in the repository.
But for binary files that are accidentally classified as text the
conversion can corrupt data.
If you recognize such corruption early you can easily fix it by
setting the conversion type explicitly in .gitattributes. Right
after committing you still have the original file in your work
tree and this file is not yet corrupted. You can explicitly tell
git that this file is binary and git will handle the file
appropriately.
Unfortunately, the desired effect of cleaning up text files with
mixed line endings and the undesired effect of corrupting binary
files cannot be distinguished. In both cases CRLFs are removed
in an irreversible way. For text files this is the right thing
to do because CRLFs are line endings, while for binary files
converting CRLFs corrupts data.
This patch adds a mechanism that can either warn the user about
an irreversible conversion or can even refuse to convert. The
mechanism is controlled by the variable core.safecrlf, with the
following values:
The default is to warn. Users are only affected by this default
if core.autocrlf is set. But the current default of git is to
leave core.autocrlf unset, so users will not see warnings unless
they deliberately chose to activate the autocrlf mechanism.
The safecrlf mechanism's details depend on the git command. The
general principles when safecrlf is active (not false) are:
- we warn/error out if files in the work tree can modified in an
irreversible way without giving the user a chance to backup the
original file.
- for read-only operations that do not modify files in the work tree
we do not not print annoying warnings.
There are exceptions. Even though...
- "git add" itself does not touch the files in the work tree, the
next checkout would, so the safety triggers;
- "git apply" to update a text file with a patch does touch the files
in the work tree, but the operation is about text files and CRLF
conversion is about fixing the line ending inconsistencies, so the
safety does not trigger;
- "git diff" itself does not touch the files in the work tree, it is
often run to inspect the changes you intend to next "git add". To
catch potential problems early, safety triggers.
The concept of a safety check was originally proposed in a similar
way by Linus Torvalds. Thanks to Dimitry Potapov for insisting
on getting the naked LF/autocrlf=true case right.
The prepare-commit-msg hook is run whenever a "fresh" commit message
is prepared, just before it is shown in the editor (if it is).
Its purpose is to modify the commit message in-place.
It takes one to three parameters. The first is the name of the file that
the commit log message. The second is the source of the commit message,
and can be: "message" (if a -m or -F option was given); "template" (if a
-t option was given or the configuration option commit.template is set);
"merge" (if the commit is a merge or a .git/MERGE_MSG file exists);
"squash" (if a .git/SQUASH_MSG file exists); or "commit", followed by
a commit SHA1 as the third parameter (if a -c, -C or --amend option
was given).
If its exit status is non-zero, git-commit will abort. The hook is
not suppressed by the --no-verify option, so it should not be used
as a replacement for the pre-commit hook.
The sample prepare-commit-msg comments out the `Conflicts:` part of
a merge's commit message; other examples are commented out, including
adding a Signed-off-by line at the bottom of the commit messsage,
that the user can then edit or discard altogether.
Signed-off-by: Paolo Bonzini <bonzini@gnu.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
This patch moves the code of run_commit, up to writing the trees, editing
the message and running the commit-msg hook to prepare_log_message. It also
renames the latter to prepare_to_commit.
This simplifies a little the code for the next patch.
When DEFAULT_GIT_TEMPLATE_DIR is specified as a relative path,
init-db made it relative to exec_path using prefix_path(), which
is wrong. prefix_path() is about a file inside the work tree.
There was a similar misuse in config.c that takes relative
ETC_GITCONFIG path. Noticed by Junio C Hamano.
We concatenate the paths manually. (prefix_filename() won't do
because it expects a prefix with a trailing '/'.)
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at> Signed-off-by: Junio C Hamano <gitster@pobox.com>
help: make 'git-help--browse' usable outside 'git-help'.
"git-help--browse" helper is to launch a browser of the user's choice
to view the HTML version of git documentation for a given command. It
used to take the name of a command, convert it to the path of the
documentation by prefixing the directory name and appending the
".html" suffix, and start the browser on the path.
This updates the division of labor between the caller in help.c and
git-help--browser helper. The helper is now responsible for launching
a browser of the user's choice on given URLs, and it is the caller's
responsibility to tell it the paths to documentation files.
This is in preparation to reuse the logic to choose user's preferred
browser in instaweb.
The helper had a provision for running it without any command name, in
which case it showed the toplevel "git(7)" documentation, but the
caller in help.c never makes such a call. The helper now exits with a
usage message when no path is given.
Signed-off-by: Christian Couder <chriscool@tuxfamily.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we process "foo/" entries in gitignore files on a system
that does not have d_type member in "struct dirent", the earlier
implementation ran lstat(2) separately when matching with
entries that came from the command line, in-tree .gitignore
files, and $GIT_DIR/info/excludes file.
This optimizes it by delaying the lstat(2) call until it becomes
absolutely necessary.
The initial idea for this change was by Jeff King, but I
optimized it further to pass pointers to around.
gitignore(5): Allow "foo/" in ignore list to match directory "foo"
A pattern "foo/" in the exclude list did not match directory
"foo", but a pattern "foo" did. This attempts to extend the
exclude mechanism so that it would while not matching a regular
file or a symbolic link "foo". In order to differentiate a
directory and non directory, this passes down the type of path
being checked to excluded() function.
A downside is that the recursive directory walk may need to run
lstat(2) more often on systems whose "struct dirent" do not give
the type of the entry; earlier it did not have to do so for an
excluded path, but we now need to figure out if a path is a
directory before deciding to exclude it. This is especially bad
because an idea similar to the earlier CE_UPTODATE optimization
to reduce number of lstat(2) calls would by definition not apply
to the codepaths involved, as (1) directories will not be
registered in the index, and (2) excluded paths will not be in
the index anyway.
An incorrect command "git mv subdir /outer/space" threw the
subdirectory to outside of the repository and then noticed that
/outer/space/subdir/ would be outside of the repository. The
error checking is backwards.
This fixes the issue by being careful about use of the return
value of get_pathspec(). Since the implementation already has
handcrafted loop to munge each path on the command line, we use
prefix_path() instead.
We would need to notice and fail if command line had a nonsense pathspec.
Earlier get_pathspec() returned all the inputs including bad ones, but
the new one issues warnings and removes offending ones from its return
value, so the callers need to be adjusted to notice it.
Additional test scripts were initially from Robin Rosenberg, further fixed.
setup: sanitize absolute and funny paths in get_pathspec()
The prefix_path() function called from get_pathspec() is
responsible for translating list of user-supplied pathspecs to
list of pathspecs that is relative to the root of the work
tree. When working inside a subdirectory, the user-supplied
pathspecs are taken to be relative to the current subdirectory.
Among special path components in pathspecs, we used to accept
and interpret only "." ("the directory", meaning a no-op) and
".." ("up one level") at the beginning. Everything else was
passed through as-is.
For example, if you are in Documentation/ directory of the
project, you can name Documentation/howto/maintain-git.txt as:
- If the pathspec is not absolute, prefix (i.e. the current
subdirectory relative to the root of the work tree, with
terminating slash, if not empty) and the pathspec is
concatenated first and used in the next step. Otherwise,
that absolute pathspec is used in the next step.
- Then special path components "." (no-op) and ".." (up one
level) are interpreted to simplify the path. It is an error
to have too many ".." to cause the intermediate result to
step outside of the input to this step.
- If the original pathspec was not absolute, the result from
the previous step is the resulting "sanitized" pathspec.
Otherwise, the result from the previous step is still
absolute, and it is an error if it does not begin with the
directory that corresponds to the root of the work tree. The
directory is stripped away from the result and is returned.
- In any case, the resulting pathspec in the array
get_pathspec() returns omit the ones that caused errors.
With this patch, the last two examples also behave as expected.
There are a few options to git-send-email to suppress the automatic
generation of 'Cc' fields: --suppress-from, and --signed-off-cc.
However, there are other times that git-send-email automatically
includes Cc'd recipients. This is not desirable for all development
environments.
Add a new option --suppress-cc, which can be specified one or more
times to list the categories of auto-cc fields that should be
suppressed. If not specified, it defaults to values to give the same
behavior as specified by --suppress-from, and --signed-off-cc. The
categories are:
self - patch sender. Same as --suppress-from.
author - patch author.
cc - cc lines mentioned in the patch.
cccmd - avoid running the cccmd.
sob - signed off by lines.
all - all non-explicit recipients
Signed-off-by: David Brown <git@davidb.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
This shares the connection between getting the remote ref list and
getting objects in the first batch. (A second connection is still used
to follow tags).
When we do not fetch objects (i.e. either ls-remote disconnects after
getting list of refs, or we decide we are already up-to-date), we
clean up the connection properly; otherwise the connection is left
open in need of cleaning up to avoid getting an error message from
the remote end when ssh is used.
Signed-off-by: Daniel Barkalow <barkalow@iabervon.org> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make error messages from cherry-pick/revert more sensible
The original "rewrite in C" did somewhat a sloppy job while
stealing code from git-write-tree.
The caller pretends as if the write_tree() function would return
an error code and being able to issue a sensible error message
itself, but write_tree() function just calls die() and never
returns an error. Worse yet, the function claims that it was
running git-write-tree (which is no longer true after
cherry-pick stole it).
Tested-by: Björn Steinbrink <B.Steinbrink@gmx.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Documentation/SubmittingPatches: discuss first then submit
This is something I've had in mind for some time. I get enough
e-mails as-is, and I suspect the workflow to get list members
involved would work better if we get the discussion concluded on
the list first before patches hit my tree (even 'next').
This new error mode allows a line to have a carriage return at the
end of the line when checking and fixing trailing whitespace errors.
Some people like to keep CRLF line ending recorded in the repository,
and still want to take advantage of the automated trailing whitespace
stripping. We still show ^M in the diff output piped to "less" to
remind them that they do have the CR at the end, but these carriage
return characters at the end are no longer flagged as errors.
git-apply --whitespace=fix: fix whitespace fuzz introduced by previous run
When you have more than one patch series, an earlier one of which
tries to introduce whitespace breakages and a later one of which
has such a new line in its context, "git-apply --whitespace=fix"
will apply and fix the whitespace breakages in the earlier one,
making the resulting file not to match the context of the later
patch.
A short demonstration is in the new test, t4125.
For example, suppose the first patch is:
diff a/hello.txt b/hello.txt
--- a/hello.txt
+++ b/hello.txt
@@ -20,3 +20,3 @@
Hello world.$
-How Are you$
-Today?$
+How are you $
+today? $
to fix broken case in the string, but it introduces unwanted
trailing whitespaces to the result (pretend you are looking at
"cat -e" output of the patch --- '$' signs are not in the patch
but are shown to make the EOL stand out). And the second patch
is to change the wording of the greeting further:
-Hello world.$
+Hello, everybody. $
How are you $
-today? $
+these days? $
If you apply the first one with --whitespace=fix, you will get
this as the result:
Hello world.$
How are you$
today?$
and this does not match the preimage of the second patch, which
demands extra whitespace after "How are you" and "today?".
This series is about teaching "git apply --whitespace=fix" to
cope with this situation better. If the patch does not apply,
it rewrites the second patch like this and retries:
-Hello world.$
+Hello, everybody.$
How are you$
-today?$
+these days?$
This is done by rewriting the preimage lines in the hunk
(i.e. the lines that begin with ' ' or '-'), using the same
whitespace fixing rules as it is using to apply the patches, so
that it can notice what it did to the previous ones in the
series.
A careful reader may notice that the first patch in the example
did not touch the "Greetings" line, so the trailing whitespace
that is in the original preimage of the second patch is not from
the series. Is rewriting this context line a problem?
If you think about it, you will realize that the reason for the
difference is because the submitter's tree was based on an
earlier version of the file that had whitespaces wrong on that
"Greetings" line, and the change that introduced the "Greetings"
line was added independently of this two-patch series to our
tree already with an earlier "git apply --whitespace=fix".
So it may appear this logic is rewriting too much, it is not
so. It is just rewriting what we would have rewritten in the
past.
builtin-apply.c: do not feed copy_wsfix() leading '+'
The "patch" parameter used to include leading '+' of an added
line in the patch, and the array was treated as 1-based. Make
it accept the contents of the line alone and simplify the code.
builtin-apply.c: simplify calling site to apply_line()
The function apply_line() changed its behaviour depending on the
ws_error_action, whitespace_error and if the input was a context.
Make its caller responsible for such checking so that we can convert
the function to copy the contents of line while fixing whitespace
breakage more easily.
We had two pointer variables pointing to the same buffer and an
integer variable used to index into its tail part that was
active (old, oldlines and oldsize for the preimage, and their
'new' counterparts for the postimage).
To help readability, use 'oldlines' as the allocated pointer,
and use 'old' as the pointer to the tail that advances while the
code builds up the contents in the buffer. The size 'oldsize'
can be computed as (old-oldines).
builtin-apply.c: mark common context lines in lineinfo structure.
This updates the way preimage and postimage in a patch hunk is
parsed and prepared for applying. By looking at image->line[n].flag,
the code can tell if it is a common context line that is the
same between the preimage and the postimage.
This matters when we actually start applying a patch with
contexts that have whitespace breakages that have already been
fixed in the target file.
builtin-apply.c: optimize match_beginning/end processing a bit.
Wnen the caller knows the hunk needs to match at the beginning
or at the end, there is no point starting from the line number
that is found in the patch and trying match with increasing
offset. The logic to find matching lines was made more line
oriented with the previous patch and this optimization is now
trivial.
This changes the way git-apply internally works to be more line
oriented. The logic to find where the patch applies with offset
used to count line numbers by always counting LF from the
beginning of the buffer, but it is simplified because we count
the line length of the target file and the preimage snippet
upfront now.
The ultimate motivation is to allow applying patches
whose preimage context has whitespace corruption that has
already been corrected in the local copy. For that purpose, we
introduce a table of line-hash that allows us to match lines
that differ only in whitespaces.
builtin-apply.c: push match-beginning/end logic down
This moves the logic to force match at the beginning and/or at
the end of the buffer to the actual function that finds the
match from its caller. This is a necessary preparation for the
next step to allow matching disregarding certain differences,
such as whitespace changes.
We probably could optimize this even more by taking advantage of
the fact that match_beginning and match_end forces the match to
be at an exact location (anchored at the beginning and/or the
end), but that's for another commit.
This restructures code to find matching location with offset
in find_offset() function, so that there is need for only one
call site of match_fragment() function. There still isn't a
change in the logic of the program.
builtin-apply.c: refactor small part that matches context
This moves three "if" conditions out of line from find_offset()
function, which is responsible for finding the matching place in
the preimage to apply the patch. There is no change in the
logic of the program.
A single signal handler is used for both SIGTERM and
SIGINT in order to clean up after an uncouth termination
of git-send-email.
In particular, the handler resets the text color (this cleanup
was already present), turns on tty echoing (in case termination
occurrs during a masked Password prompt), and informs the user
of of any temporary files created by --compose.
Signed-off-by: Michael Witten <mfwitten@mit.edu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Whilst convenient, it is most unwise to record passwords
in any place but one's brain. Moreover, it is especially
foolish to store them in configuration files, even with
access permissions set accordingly.
git-send-email has been amended, so that if it detects
an smtp username without a password, it promptly prompts
for the password and masks the input for privacy.
Furthermore, the argument to --smtp-pass has been rendered
optional.
The documentation has been updated to reflect these changes.
Signed-off-by: Michael Witten <mfwitten@mit.edu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Jakub Narebski <jnareb@gmail.com> wrote Sun, Feb 03, 2008:
> Junio C Hamano wrote:
> > Jakub Narebski <jnareb@gmail.com> writes:
> >
> > [From] http://thread.gmane.org/gmane.comp.version-control.git/53457/focus=53458
> Julian Phillips:
> > Are you using docbook xsl 1.72? There are known problems building the
> > manpages with that version. 1.71 works, and 1.73 should work when it get
> > released.
I was able to solve this problem with this patch, which adds a XSL file
used specifically for DOCBOOK_XSL_172=YesPlease and where dots and
backslashes are escaped properly so they won't be substituted to the
wrong thing further down the "DocBook XSL pipeline". Doing the escaping
in the existing callout.xsl breaks v1.70.1. Hopefully v1.73 will end
this part of the manpage nightmare.
When DEFAULT_GIT_TEMPLATE_DIR is specified as a relative path,
init-db made it relative to exec_path using prefix_path(), which
is wrong. prefix_path() is about a file inside the work tree.
There was a similar misuse in config.c that takes relative
ETC_GITCONFIG path.
A convenience function prefix_filename() can concatenate two paths
to form a path that points at somewhere outside the work tree.
Use it in these codepaths instead.
parse_object_buffer: don't ignore errors from the object specific parsing functions
In the case of an malformed object, the object specific parsing functions
would return an error, which is currently ignored. The object can be partial
initialized in this case.
This patch make parse_object_buffer propagate such errors.
Signed-off-by: Martin Koegler <mkoegler@auto.tuwien.ac.at> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-fsck: report missing author/commit line in a commit as an error
A zero commit date could be caused by:
* a missing author line
* a missing commiter line
* a malformed email address in the commiter line
* a malformed commit date
Simply reporting it as zero commit date is missleading.
Additionally, it upgrades the message to an error (instead of an printf).
Signed-off-by: Martin Koegler <mkoegler@auto.tuwien.ac.at> Signed-off-by: Junio C Hamano <gitster@pobox.com>