gitweb: Refactor feed generation, make output prettier, add Atom feed
Add support for more modern Atom web feed format. Both RSS and Atom
feeds are generated by git_feed subroutine to avoid code duplication;
git_rss and git_atom are thin wrappers around git_feed. Add links to
Atom feed in HTML header and in page footer (but not in OPML; we
should use APP, Atom Publishing Proptocol instead).
Allow for feed generation for branches other than current (HEAD)
branch, and for generation of feeds for file or directory history.
Do not use "pre ${\sub_returning_scalar(...)} post" trick, but join
strings instead: "pre " . sub_returning_scalar(...) . " post".
Use href(-full=>1, ...) instead of hand-crafting gitweb urls.
Make output prettier:
* Use title similar to the title of web page
* Use project description (if exists) for description/subtitle
* Do not add anything (committer name, commit date) to feed entry title
* Wrap the commit message in <pre>
* Make file names into an unordered list
* Add links (diff, conditional blame, history) to the file list.
In addition to the above points, the attached patch emits a
Last-Changed: HTTP response header field, and doesn't compute the feed
body if the HTTP request type was HEAD. This helps keep the web server
load down for well-behaved feed readers that check if the feed needs
updating.
If browser (feed reader) sent Accept: header, and it prefers 'text/xml' type
to 'application/rss+xml' (in the case of RSS feed) or 'application/atom+xml'
(in the case of Atom feed), then use 'text/xml' as content type.
Both RSS and Atom feeds validate at http://feedvalidator.org
and at http://validator.w3.org/feed/
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Andreas Fuchs <asf@boinkor.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Add an option to href() to return full URL
href subroutine by default generates absolute URL (generated using
CGI::url(-absolute=>1), and saved in $my_uri) using $my_uri as base;
add an option to generate full URL using $my_url as base.
New feature usage: href(..., -full=>1)
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: New improved formatting of chunk header in diff
If we have provided enough info, and diff is not combined diff,
and if provided diff line is chunk header, then:
* split chunk header into .chunk_info and .section span elements,
first containing proper chunk header, second section heading
(aka. which function), for separate styling: the proper chunk
header is on non-white background, section heading part uses
slightly lighter color.
* hyperlink from-file-range to starting line of from-file, if file
was not created.
* hyperlink to-file-range to starting line of to-file, if file
was not deleted.
Links are of invisible variety (and "list" class).
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Default to $hash_base or HEAD for $hash in "commit" and "commitdiff"
Set $hash parameter to $hash_base || "HEAD" if it is not set (if it is
not true to be more exact). This allows [hand-edited] URLs with 'action'
"commit" or "commitdiff" but without 'hash' parameter.
If there is 'h' (hash) parameter provided, then gitweb tries
to use this. HEAD is used _only_ if nether hash, nor hash_base
are provided, i.e. for URL like below
URL?p=project.git;a=commit
i.e. without neither 'h' nor 'hb'.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Buffer diff header to deal with split patches + git_patchset_body refactoring
There are some cases when one line from "raw" git-diff output (raw format)
corresponds to more than one patch in the patchset git-diff output. To deal
with this buffer git diff header and extended diff header (everything up to
actual patch) to check from information from "index <hash>..<hash>" extended
header line if the patch corresponds to the same or next difftree raw line.
This could also be used to gather information needed for hyperlinking, and
used for printing gitweb quoted filenames, from extended diff header instead
of raw git-diff output.
While at it, refactor git_patchset_body subroutine from the event-driven,
AWK-like state-machine parsing to sequential parsing: for each patch
parse (and output) git diff header, parse extended diff header, parse two-line
from-file/to-file diff header, parse patch itself; patch ends with the end
of input [file] or the line matching m/^diff /.
For better understanding the code, there were added assertions in the
comments a la Carp::Assert module. Just in case there is commented out code
dealing with unexpected end of input (should not happen, hence commented
out).
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
As discussed on git mailing list let's teach the reader about
the possiblity to have automatically signed off the commit running
the git-commit -s command
Signed-off-by: Paolo Ciarrocchi <paolo.ciarrocchi@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
do_for_each_ref: perform the same sanity check for leftovers.
An earlier commit b37a562a added a check to see if the ref
points at a valid object (as a part of 'negative ref' support
which we currently do not use), but did so only while iterating
over both packed and loose refs, and forgot to apply the same
check while iterating over the remaining ones.
We might want to replace the "if null then omit it" check with
"eh --- what business does a 0{40} value have here?" complaint
later since we currently do not use negative refs, but that is
a separate issue.
For one, the documentation invalidly claimed that the paths have to be
absolute when that's not the case and in fact there is a very valid reason
not to use absolute paths (documented the reason as well).
Signed-off-by: Petr Baudis <pasky@suse.cz> Signed-off-by: Junio C Hamano <junkio@cox.net>
This patch fixes a sparse warning about inaccurate_eof being a
"dubious one-bit signed bitfield", makes three more binary
variables members of this (now unsigned) bitfield and adds a
short comment to indicate the nature of two ternary variables.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx> Signed-off-by: Junio C Hamano <junkio@cox.net>
Unfortunately, git-for-each-refs is currently unusable for peeking into tag
comments, since it uses freed pointers, so it just prints out all sort of
garbage.
This makes it strdup() contents and body values.
Signed-off-by: Petr Baudis <pasky@suse.cz> Signed-off-by: Junio C Hamano <junkio@cox.net>
Ok, this is a _really_ stupid case, and I don't think it matters, but hey,
we should never SIGSEGV.
Steps to reproduce:
mkdir duh
cd duh
git init-db
git-fmt-merge-msg < /dev/null
will cause a SIGSEGV in cmd_fmt_merge_msg(), because we're doing a
strncmp() with a NULL current_branch.
And yeah, it's an insane schenario, and no, it doesn't really matter. The
only reason I noticed was that a broken version of my "git pull" into an
empty directory would cause this.
This silly patch just replaces the SIGSEGV with a controlled exit with an
error message.
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
We used to complain that we cannot merge anything we fetched
with a local branch that does not exist yet. Just treat the
case as a natural extension of fast forwarding and make the
local branch'es tip point at the same commit we just fetched.
After all an empty repository without an initial commit is an
ancestor of any commit.
[jc: I added a trivial test. We've become sloppy but we should
stick to the discipline of covering new behaviour with new
tests. ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Run "git repack -a -d" once more at end, if there's 1MB or more of not-packed data.
Although I converted upstream coreutils to git last month, I just
reconverted coreutils once again, as a test, and ended up with a
git repository of about 130MB (contrast with my packed git repo of
size 52MB). That was because there were a lot of commits (but < 1024)
after the final automatic "git-repack -a -d".
Running a final
git-repack -a -d && git-prune-packed
cut the final repository size down to the expected size.
So this looks like an easy way to improve git-cvsimport.
Just run "git repack ..." at the end if there's more than
some reasonable amount of not-packed data.
My choice of 1MB is a little arbitrarily. I wouldn't mind missing
the minimal repo size by 1MB. At the other end of the spectrum,
it's probably not worthwhile to pack everything when the total
repository size is less than 1MB.
Here's the patch:
Signed-off-by: Jim Meyering <jim@meyering.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
Seek back to current filepos when mmap()ing with NO_MMAP
"git-index-pack --fix-thin" relies on mmap() not changing the current
file position (otherwise the pack will be corrupted when writing the
final SHA1). Meet that expectation.
Signed-off-by: Johannes Schindelin <Johannes.Schindelin@gmx.de> Signed-off-by: Junio C Hamano <junkio@cox.net>
* maint:
Rework cvsexportcommit to handle binary files for all cases.
Catch errors when writing an index that contains invalid objects.
test-lib.sh: A command dying due to a signal is an unexpected failure.
git-update-index(1): fix use of quoting in section title
Catch errors when writing an index that contains invalid objects.
If git-write-index is called without --missing-ok, it reports invalid
objects that it finds in the index. But without this patch it dies
right away or may run into an infinite loop.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at> Signed-off-by: Junio C Hamano <junkio@cox.net>
When path-list-insert is called on an existing path, it returned an
unrelated element in the list. Luckily most of the callers are
ignoring the return value, but merge-recursive uses it at three places
and this would have resulted in a bogus rename detection.
* maint:
git-rebase: Use --ignore-if-in-upstream option when executing git-format-patch.
git-svn: fix dcommit losing changes when out-of-date from svn
git-svn: don't die on rebuild when --upgrade is specified
git-svn: avoid printing filenames of files we're not tracking
git-svn: fix dcommit losing changes when out-of-date from svn
There was a bug in dcommit (and commit-diff) which caused deltas
to be generated against the latest version of the changed file
in a repository, and not the revision we are diffing (the tree)
against locally.
This bug can cause recent changes to the svn repository to be
silently clobbered by git-svn if our repository is out-of-date.
Thanks to Steven Grimm for noticing the bug.
The (few) people using the commit-diff command are now required
to use the -r/--revision argument. dcommit usage is unchanged.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-svn: don't die on rebuild when --upgrade is specified
--copy-remote and --upgrade are rarely (never?) used together,
so if --copy-remote is specified, that means the user really
wanted to copy the remote ref, and we should fail if that fails.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
Nicer error messages in case saving an object to db goes wrong
Currently the error e.g. when pushing to a read-only repository is quite
confusing, this attempts to clean it up, unifies error reporting between
various object writers and uses error() on couple more places.
Signed-off-by: Petr Baudis <pasky@suse.cz> Signed-off-by: Junio C Hamano <junkio@cox.net>
The same change as the previous. It is rather sad that commit log
message parser gives list of chomped lines while tag message parser
gives unchomped ones.
gitweb: protect blob and diff output lines from controls.
This revealed that the output from blame and tag was not chomped
properly and was relying on HTML output not noticing that extra
whitespace that resulted from the newline, which was also fixed.
Just make it take over blame's place. Documentation and command
have all stopped mentioning "git-pickaxe". The built-in synonym
is left in the command table, so you can still say "git pickaxe",
but it probably is a good idea to retire it as well.
Otherwise, commit log template would get the remainder of the
filename start on a new line unquoted and the log gets messed
up.
I initially considered using the full quote_c_style(), but the
output from the command is primarily for human consumption so
chose to leave other control characters and bytes with high-bits
unmolested.
Replace "gitweb diff header" with its full sha1 of blobs and replace
it by "git diff" header and extended diff header. Change also somewhat
highlighting of diffs.
Added `file_type_long' subroutine to convert file mode in octal to
file type description (only for file modes which used by git).
Changes:
* "gitweb diff header" which looked for example like below:
file:_<sha1 before>_ -> file:_<sha1 after>_
where 'file' is file type and '<sha1>' is full sha1 of blob is
changed to
diff --git _a/<file before>_ _b/<file after>_
In both cases links are visible and use default link style. If file
is added, a/<file> is not hyperlinked. If file is deleted, b/<file>
is not hyperlinked.
* there is added "extended diff header", with <path> and <hash>
hyperlinked (and <hash> shortened to 7 characters), and <mode>
explained: '<mode>' is extended to '<mode> (<file type description>)',
where added text is slightly lighter to easy distinguish that it
was added (and it is difference from git-diff output).
* from-file/to-file two-line header lines have slightly darker color
than removed/added lines.
* chunk header has now delicate line above for easier finding chunk
boundary, and top margin of 2px, both barely visible.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Use character or octal escape codes (and add span.cntrl) in esc_path
Instead of simply hiding control characters in esc_path by replacing
them with '?', use Character Escape Codes (CEC) i.e. alphabetic
backslash sequences like those found in C programming language and
many other languages influenced by it, such as Java and Perl. If
control characted doesn't have corresponding character escape code,
use octal char sequence to escape it.
Alternatively, controls can be replaced with Unicode Control
Pictures U+2400 - U+243F (9216 - 9279), the Unicode characters
reserved for representing control characters when it is
necessary to print or display them.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Better git-unquoting and gitweb-quoting of pathnames
Extend unquote subroutine, which unquotes quoted and escaped filenames
which git may return, to deal not only with octal char sequence
quoting, but also quoting ordinary characters including '\"' and '\\'
which are respectively quoted '"' and '\', and to deal also with
C escape sequences including '\t' for TAB and '\n' for LF.
Add esc_path subroutine for gitweb quoting and HTML escaping filenames
(currently it does equivalent of ls' --hide-control-chars, which means
showing undisplayable characters (including '\n' and '\t') as '?'
(question mark) character, and use 'span' element with cntrl CSS class
to help rendering them differently.
Convert gitweb to use esc_path correctly to print pathnames.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
A forked project is defined to be $projname/$forkname.git for
$projname.git; the code did not check this correctly and mistook
$projname/.git to be a fork of itself. This minimally fixes the
breakage.
Also forks were not checked when index.aux file was in use.
Listing the forked ones in index.aux would show them also on the
toplevel index which may go against the hierarchical nature of
forks, but again this is a minimal fix to whip it in a better
shape suitable to be in the 'master' branch.
Apparently this code was never tested without "forks". check-feature
returns a one-element list (0) when disabled, and assigning that to a
scalar variable made it to be called in a scalar context, which meant
my $check_forks = gitweb_check_feature("forks") were always 1!
* jc/read-tree:
t6022: ignoring untracked files by merge-recursive when they do not matter
merge-recursive: adjust to loosened "working file clobbered" check
merge-recursive: make a few functions static.
merge-recursive: use abbreviated commit object name.
merge: loosen overcautious "working file will be lost" check.
* np/index-pack:
remove .keep pack lock files when done with refs update
have index-pack create .keep file more carefully
improve fetch-pack's handling of kept packs
git-fetch can use both --thin and --keep with fetch-pack now
Teach receive-pack how to keep pack files based on object count.
Allow pack header preprocessing before unpack-objects/index-pack.
Remove unused variable in receive-pack.
Revert "send-pack --keep: do not explode into loose objects on the receiving end."
missing small substitution
Teach git-index-pack how to keep a pack file.
Only repack active packs by skipping over kept packs.
Allow short pack names to git-pack-objects --unpacked=.
send-pack --keep: do not explode into loose objects on the receiving end.
index-pack: minor fixes to comment and function name
enhance clone and fetch -k experience
mimic unpack-objects when --stdin is used with index-pack
add progress status to index-pack
make index-pack able to complete thin packs.
enable index-pack streaming capability
* maint:
Documentation: Transplanting branch with git-rebase --onto
merge-recursive implicitely depends on trust_executable_bit
adjust_shared_perm: chmod() only when needed.
Fix git-runstatus for repositories containing a file named HEAD
Documentation: Transplanting branch with git-rebase --onto
Added example of transplantig feature branch from one development
branch (for example "next") into the other development branch (for
example "master").
[jc: talking Carl's advice this contains both examples sent to
the list by Jakub in his original message.]
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
When widening permission for files and directories in a 'shared'
repository for a user with inappropriate umask() setting for
shared work, make sure we call chmod() only when we actually
need to.
Fix git-runstatus for repositories containing a file named HEAD
The wt_status_print_updated() and wt_status_print_untracked() routines
call setup_revisions() with 'HEAD' being the reference to the tip of the
current branch. However, setup_revisions() gets confused if the branch
also contains a file named 'HEAD' resulting in a fatal error.
Instead, don't pass an argv to setup_revisions() at all; simply give it no
arguments, and make 'HEAD' the default revision.
Bug noticed by Rocco Rutte <pdmef@gmx.net>.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
Remove more sed invocations from within bash completion.
This change removes between 1 and 4 sed invocations per completion
entered by the user. In the case of cat-file the 4 invocations per
completion can take a while on Cygwin; running these replacements
directly within bash saves some time for the end user.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Support bash completion on symmetric difference operator.
Now that log, whatchanged, rev-list, etc. support the symmetric
difference operator '...' we should provide bash completion for it
just like we do for '..'.
While we are at it we can remove two sed invocations during the
interactive prompt and replace them with internal bash operations.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Take --git-dir into consideration during bash completion.
If the user has setup a command line of "git --git-dir=baz" then
anything we complete must be performed within the scope of "baz"
and not the current working directory.
This is useful with commands such as "git --git-dir=git.git log m"
to complete out "master" and view the log for the master branch of
the git.git repository. As a nice side effect this also works for
aliases within the target repository, just as git would honor them.
Unfortunately because we still examine arguments by absolute position
in most of the more complex commands (e.g. git push) using --git-dir
with those commands will probably still cause completion to fail.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Bash completion support for remotes in .git/config.
Now that Git natively supports remote specifications within the
config file such as:
[remote "origin"]
url = ...
we should provide bash completion support "out of the box" for
these remotes, just like we do for the .git/remotes directory.
Also cleaned up the __git_aliases expansion to use the same form
of querying and filtering repo-config as this saves two fork/execs
in the middle of a user prompted completion. Finally also forced
the variable 'word' to be local within __git_aliased_command.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Added missing completions for show-branch and merge-base.
The show-branch and merge-base commands were partially supported
when it came to bash completions as they were only specified in
one form another. Now we specify them in both forms.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-pickaxe: optimize by avoiding repeated read_sha1_file().
It turns out that pickaxe reads the same blob repeatedly while
blame can reuse the blob already read for the parent when
handling a child commit when it's parent's turn to pass its
blame to the grandparent. Have a cache in the origin structure
to keep the blob there, which will be garbage collected when the
origin loses the last reference to it.
* maint:
Remove unsupported C99 style struct initializers in git-archive.
Remove SIMPLE_PROGRAMS and make git-daemon a normal program.
Use ULONG_MAX rather than implicit cast of -1.
Remove unsupported C99 style struct initializers in git-archive.
At least one older version of the Solaris C compiler doesn't support
the newer C99 style struct initializers. To allow Git to compile
on those systems use an archive description struct which is easier
to initialize without the C99 struct initializer syntax.
Also since the archives array is not used by anyone other than
archive.c we can make it static.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
Remove SIMPLE_PROGRAMS and make git-daemon a normal program.
Some platforms (Solaris in particular) appear to require -lz as
part of the link line for git-daemon, due to it linking against
sha1_file.o and that module requiring inflate/deflate support.
So its time to retire SIMPLE_PROGRAMS and move its last remaining
member into the standard PROGRAMS list, allowing it to link against
all libraries used by the rest of Git.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
At least one (older) version of the Solaris C compiler won't allow
'unsigned long x = -1' without explicitly casting -1 to a type of
unsigned long. So instead use ULONG_MAX, which is really the
correct constant anyway.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-svn: don't die on rebuild when --upgrade is specified
--copy-remote and --upgrade are rarely (never?) used together,
so if --copy-remote is specified, that means the user really
wanted to copy the remote ref, and we should fail if that fails.
Signed-off-by: Eric Wong <normalperson@yhbt.net> Signed-off-by: Junio C Hamano <junkio@cox.net>
When we introduced the cached origin per commit, we gave up proper
garbage collecting because it meant that commits hold onto their
cached copy. There is no need to do so.
On Cygwin a user might complete the new git-branch builtin as
git-branch.exe, at which point bash requires a new completion
registration for the command.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-pickaxe: re-scan the blob after making progress with -C
The reason to do this is the same as in the previous change for
line copy detection within the same file (-M).
Also this fixes -C and -C -C (aka find-copies-harder) logic; in
this application we are not interested in the similarity
matching diffcore-rename makes, because we are only interested
in scanning files that were modified, or in the case of -C -C,
scanning all files in the parent and we want to do that
ourselves.
git-pickaxe: re-scan the blob after making progress with -M
Otherwise we would miss copied lines that are contained in the
parts before or after the part that we find after splitting the
blame_entry (i.e. split[0] and split[2]).
We deduced a GNU diff output that does not use /dev/null convention
as creation (deletion) diff correctly by looking at the lack of context
and deleted lines (added lines), but forgot to reset the new (old) name
field properly.
This was a regression when we added a workaround for --unified=0 insanity.