Remove git_diff_print subroutine, used to print diff in previous
versions of "diff" actions, namely git_commitdiff,
git_commitdiff_plain, git_blobdiff, git_blobdiff_plain.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: git_blobdiff_plain is git_blobdiff('plain')
git_blobdiff and git_blobdiff_plain are now collapsed into one
subroutine git_blobdiff, with format (currently 'html' which is
default format corresponding to git_blobdiff, and 'plain'
corresponding to git_blobdiff_plain) specified in argument.
blobdiff_plain format is now generated either by git-diff-tree
or by git-diff. Added X-Git-Url: header. From-file and to-file name
in header are corrected.
Note that for now commitdiff_plain does not detect renames
and copying, while blobdiff_plain does.
While at it, set expires to "+1d" for non-textual hash ids.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Use git-diff-tree or git-diff patch output for blobdiff
This is second part of removing gitweb dependency on external
diff (used in git_diff_print).
Get rid of git_diff_print invocation in git_blobdiff, and use either
git-diff-tree (when both hash_base and hash_parent_base are provided)
patch format or git-diff patch format (when only hash and hash_parent
are provided) for output.
Supported URI schemes, and output formats:
* New URI scheme: both hash_base and hash_parent_base (trees-ish
containing blobs versions we want to compare) are provided.
Also either filename is provided, or hash (of blob) is provided
(we try to find filename then).
For this scheme we have copying and renames detection, mode changes,
file types etc., and information extended diff header is correct.
* Old URI scheme: hash_parent_base is not provided, we use hash and
hash_parent to directly compare blobs using git-diff. If no filename
is given, blobs hashes are used in place of filenames.
This scheme has always "blob" as file type, it cannot detect mode
changes, and we rely on CGI parameters to provide name of the file.
Added git_to_hash subroutine, which transforms symbolic name or list
of symbolic name to hash or list of hashes using git-rev-parse.
To have "blob" instead of "unknown" (or "file" regardless of the type)
in "gitweb diff header" for legacy scheme, file_type function now
returns its argument if it is not octal string.
Added support for fake "2" status code in git_patchset_body. Such code
is generated by git_blobdiff in legacy scheme case.
ATTENTION: The order of arguments (operands) to git-diff is reversed
(sic!) to have correct diff in the legacy (no hash_parent_base) case.
$hash_parent, $hash ordering is commented out, as it gives reversed
patch (at least for git version 1.4.1.1) as compared to output in new
scheme and output of older gitweb version.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Add invisible hyperlink to from-file/to-file diff header
Change replacing hashes as from-file/to-file with filenames from
difftree to adding invisible (except underlining on hover/mouseover)
hyperlink to from-file/to-file blob. /dev/null as from-file or
to-file is not changed (is not hyperlinked).
This makes two-file from-file/to-file unified diff header parsing in
git_patchset_body more generic, and not only for legacy blobdiffs.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Parse two-line from-file/to-file diff header in git_patchset_body
Parse two-line from-file/to-file unified diff header in
git_patchset_body directly, instead of leaving pretty-printing to
format_diff_line function. Hashes as from-file/to-file are replaced
by proper from-file and to-file names (from $diffinfo); in the future
we can put hyperlinks there. This makes possible to do blobdiff with
only blobs hashes.
The lines in two-line unified diff header have now class "from_file"
and "to_file"; the style is chosen to match previous output (classes
"rem" and "add" because of '-' and '+' as first character of patch
line).
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Add support for hash_parent_base parameter for blobdiffs
Add support for hash_parent_base in input validation part and in
href() function. Add proper hash_parent_base to all calls to blobdiff
and blobdiff_plain action URLs. Use hash_parent_base as hash_base for
blobs of hash_parent.
To be used in future rewrite of git_blobdiff and git_blobdiff_plain.
While at it, move project before action in ordering CGI parameters in
href().
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Adds git_get_following_references function, based on code which was
used in git_commitdiff_plain to generate X-Git-Tag: header,
and companion git_get_preceding_references function.
Both functions return array of all references of given type (as
returned by git_get_references) following/preceding given commit in
array (list) context, and last following/first preceding ref in scalar
context.
Stripping ref (list of refs) of "$type/" (e.g. "tags/") is left to
caller.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Change output of patch(set) in git_commitdiff from slurping whole diff
in @patchset array before processing, to passing file descriptor to
git_patchset_body.
Advantages: faster, incremental output, smaller memory footprint.
Disadvantages: cannot react when there is error during closing file
descriptor.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Show information about incomplete lines in commitdiff
In format_diff_line, instead of skipping errors/incomplete lines,
for example
"\ No newline at end of file"
in HTML pretty-printing of diff, use "incomplete" class for div.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Use git-diff-tree patch output for commitdiff
Get rid of git_diff_print invocation in git_commitdiff and therefore
external diff (/usr/bin/diff) invocation, and use only git-diff-tree
to generate patch.
git_commitdiff and git_commitdiff_plain are collapsed into one
subroutine git_commitdiff, with format (currently 'html' which is
default format corresponding to git_commitdiff, and 'plain'
corresponding to git_commitdiff_plain) specified in argument.
Separate patch (diff) pretty-printing into git_patchset_body.
It is used in git_commitdiff.
Separate patch (diff) line formatting from git_diff_print into
format_diff_line function. It is used in git_patchset_body.
While at it, add $hash parameter to git_difftree_body, according to
rule that inner functions should use parameter passing, and not global
variables.
CHANGES TO OUTPUT:
* "commitdiff" now products patches with renaming and copying
detection (git-diff-tree is invoked with -M and -C options).
Empty patches (mode changes and pure renames and copying)
are not written currently. Former version broke renaming and
copying, and didn't notice mode changes, like this version.
* "commitdiff" output is now divided into several div elements
of class "log", "patchset" and "patch".
* "commitdiff_plain" now only generates X-Git-Tag: line only if there
is tag pointing to the current commit. Former version which wrote
first tag following current commit was broken[*1*]; besides we are
interested rather in tags _preceding_ the commit, and _heads_
following the commit. X-Git-Url: now is current URL; former version
tried[*2*] to output URL to HTML version of commitdiff.
* "commitdiff_plain" is generated by git-diff-tree, and has therefore
has git specific extensions to diff format: "git diff" header and
optional extended header lines.
FOOTNOTES
[*1*] First it generated rev-list starting from HEAD even if hash_base
parameter was set, second it wasn't corrected according to changes
made in git_get_references (formerly read_info_ref) output, third even
for older version of read_info_ref output it didn't work for multiple
tags pointing to the current commit (rare).
[*2*] It wrote URL for commitdiff without hash_parent, which produces
diff to first parent and is not the same as current diff if it is diff
of merge commit to non-first parent.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Use list continuation to have better wrapping. This accounts for most of
the changes because it reindents a lot of text without applying other
changes.
Use cross-referencing for interlinking and the gitlink macro for pointing
to other tools in the git suite.
Signed-off-by: Jonas Fonseca <fonseca@diku.dk> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitk(1): expand the manpage to look less like a template
Add a short description and document a few selected options additionally to
the different "entities" in the standard calling convention. Advertise
other git repository browsers. Lastly, climb Mount Ego.
Signed-off-by: Jonas Fonseca <fonseca@diku.dk> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-blame(1): mention options in the synopsis and advertise pickaxe
Inspired by the cvs annotate documentation improve and expand the man page
to also mention the limitations of file annotations. Since people coming
from the SVN/CVS world might first look here, also briefly advertise how
the pickaxe interface makes it easy to go beyond these limitation.
Signed-off-by: Jonas Fonseca <fonseca@diku.dk> Signed-off-by: Junio C Hamano <junkio@cox.net>
This abstracts away the size of the hash values when copying them
from memory location to memory location, much as the introduction
of hashcmp abstracted away hash value comparsion.
A few call sites were using char* rather than unsigned char* so
I added the cast rather than open hashcpy to be void*. This is a
reasonable tradeoff as most call sites already use unsigned char*
and the existing hashcmp is also declared to be unsigned char*.
[jc: this is a follow-up patch for merge-recursive.c which is
not in "master" yet. The original was sent-in for "next" so
I splitted it out. ]
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
* gl/cleanup: (160 commits)
Convert memset(hash,0,20) to hashclr(hash).
Convert memcpy(a,b,20) to hashcpy(a,b).
Fix a comparison bug in diff-delta.c
git-send-email: Don't set author_not_sender from Cc: lines
Remove unnecessary forward declaration of unpack_entry.
Verify we know how to read a pack before trying to using it.
Add write_or_die(), a helper function
Axe the last ent
builtin-mv: readability patch
git-mv: fix off-by-one error
git-mv: special case destination "."
builtin-mv: readability patch
Indentation fix.
Do not use memcmp(sha1_1, sha1_2, 20) with hardcoded length.
gitweb: Uniquify version info output, add meta generator in page header
Be nicer if git executable is not installed
builtin-grep: remove unused debugging cruft.
gitweb: Add support for per project git URLs
[PATCH] git-mv: add more path normalization
Remove the "delay writing to avoid runtime penalty of racy-git avoidance"
...
This abstracts away the size of the hash values when copying them
from memory location to memory location, much as the introduction
of hashcmp abstracted away hash value comparsion.
A few call sites were using char* rather than unsigned char* so
I added the cast rather than open hashcpy to be void*. This is a
reasonable tradeoff as most call sites already use unsigned char*
and the existing hashcmp is also declared to be unsigned char*.
[jc: Splitted the patch to "master" part, to be followed by a
patch for merge-recursive.c which is not in "master" yet.
Fixed the cast in the latter hunk to combine-diff.c which was
wrong in the original.
Also converted ones left-over in combine-diff.c, diff-lib.c and
upload-pack.c ]
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
(1 << i) < hspace is compared in the `int` space rather that in the
unsigned one. the result will be wrong if hspace is between 0x40000000
and 0x80000000.
Signed-off-by: Pierre Habouzit <madcoder@debian.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
git-send-email: Don't set author_not_sender from Cc: lines
When an mbox-style patch contains a Cc: line in the header,
git-send-email will check the address against the sender specified
on the command line. If they don't match, sender_not_author will
be set to the address obtained from the Cc line.
When this happens, git-send-email inserts a From: line at the
beginning of the message body with the address obtained from the
Cc line in the header, and the sender might be accused of forging
patch authors.
This patch fixes this by only updating sender_not_author when
processing From: lines, not when processing Cc: lines.
Signed-off-by: Haavard Skinnemoen <hskinnemoen@atmel.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Drop the href() params which keys are not in %mapping
If someone would enter parameter name incorrectly, and some key of
%params is not found in %mapping hash, the parameter is now
ignored. Change introduced by Martin Waitz in commit 756d2f064b2419fcdf9cd9c851f352e2a4f75103
tried to do that, but it left empty value and there was doubled ";;"
in returned string.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Route rest of action subroutines through %actions
Route rest of action subroutines, namely git_project_list and git_opml
(both of which doesn't need $project) through %actions hash.
This has disadvantage that all parameters are read and validated;
earlier git_opml was called as soon as $action was parsed and
validated, git_project_list was called as soon as $project was parsed
and validated. This has advantage that all action dispatch is grouped
in one place.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
This patch tries (but no too hard) to fit gitweb source in 80 columns,
for 2 columns wide tabs, and indent and align source for better
readibility.
While at it added comment to 'snapshot' entry defaults for %feature
hash, corrected "blobl" action in git_blame2 and git_blame to "blob",
key of argument to $cgi->a from 'class' to '-class'.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
"a.list" being "bold", makes a myriad of things shown by
gitweb in bold font-weight, which is a regression from
pre-17d07443188909ef5f8b8c24043cb6d9fef51bca behavior.
The fix is to add "subject" class and use this class
to replace pre-format_subject_html formatting of subject
(comment) via using (or not) <b>...</b> element. This
should go back to the pre-17d0744318... style.
Regression noticed by Luben Tuikov.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Verify we know how to read a pack before trying to using it.
If the pack format were to ever change or be extended in the future
there is no assurance that just because the pack file lives in
objects/pack and doesn't end in .idx that we can read and decompress
its contents properly.
If we encounter what we think is a pack file and it isn't or we don't
recognize its version then die and suggest to the user that they
upgrade to a newer version of GIT which can handle that pack file.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
The little helper write_or_die() won't come back with bad news about
full disks or broken pipes. It either succeeds or terminates the
program, making additional error handling unnecessary.
This patch adds the new function and uses it to replace two similar
ones (the one in tar-tree originally has been copied from cat-file
btw.). I chose to add the fd parameter which both lacked to make
write_or_die() just as flexible as write() and thus suitable for
lib-ification.
There is a regression: error messages emitted by this function don't
show the program name, while the replaced two functions did. That's
acceptable, I think; a lot of other functions do the same.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx> Signed-off-by: Junio C Hamano <junkio@cox.net>
In the name of Standardization, this cleanses the last usage string of
mystical creatures. But they still dwell deep within the source and in
some debug messages, it is said.
Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: Use parse_difftree_raw_line in git_difftree_body
Use newly introduced parse_difftree_raw_line function in the
git_difftree_body subroutine. While at it correct error in
parse_difftree_raw_line (unquote is unprototyped function), and
add comment explaining this function.
It also refactors git_difftree_body somewhat, and tries to fit
it in 80 columns.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
This adds blame and snapshot to the feature associative array. This
also helps in enabling or disabling these features via GITWEB_CONFIG
and overriding if allowed via project specfic config.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
[jc: when I applied the patch I misread RFC 2616 which mildly
recommended against using the name "gzip", which was there only
for a historical reason. This fixes the mistake up and uses
the content-encoding "x-gzip" again.]
gitweb: bugfix: git_print_page_path() needs the hash base
If a file F exists in branch B, but doesn't exist
in master branch, then blob_plain needs the hash base in
order to properly get the file. The hash base is passed
on symbolically so we still preserve the "latest" quality
of the link presented by git_print_page_path().
Signed-off-by: Luben Tuikov <ltuikov@yahoo.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
This adds snapshort support in gitweb. To enable one need to
set gitweb.snapshot = true in the config file. Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@hp.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
Separate pretty-printing commit message (comment) into git_print_log
and git_print_simplified_log subroutines. As of now the former is used
in git_commit, the latter in git_log and git_commitdiff.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
The project list now uses several common header / footer generation functions.
These functions only check for "defined $project", but when PATH_INFO just
contains a "/" (which is often generated by web servers), then this test
fails.
Now explicitly undef $project if there is none so that the tests in other
gitweb parts work again.
Signed-off-by: Martin Waitz <tali@admingilde.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
If the webserver is configured to use gitweb even for the root directory
of the site, then my_uri is empty which leads to a non-functional home
link. Fix that by defaulting to "/" in this case.
Signed-off-by: Martin Waitz <tali@admingilde.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
With the new flag "--reject", hunks that do not apply are sent to
the standard output, and the usable hunks are applied. The command
itself exits with non-zero status when this happens, so that the
user or wrapper can take notice and sort the remaining mess out.
This patch avoids problems if vc-git.el is installed and activated, but
the git executable is not available, for example
http://list-archive.xemacs.org/xemacs-beta/200608/msg00062.html
Signed-off-by: Ville Skyttä <scop@xemacs.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
* jc/racy:
Remove the "delay writing to avoid runtime penalty of racy-git avoidance"
Add check program "git-check-racy"
Documentation/technical/racy-git.txt
avoid nanosleep(2)
It is now possible for project to have individual clone/fetch URLs.
They are provided in new file 'cloneurl' added below project's
$GIT_DIR directory.
If there is no cloneurl file, concatenation of git base URLs with
project name is used.
This is merge of Jakub Narebski and David Rientjes
gitweb: Show project's git URL on summary page
with Aneesh Kumar
gitweb: Add support for cloneurl.
gitweb: Support multiple clone urls
patches.
Signed-off-by: Jakub Narebski <jnareb@gmail.com> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@gmail.com> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitk: Improve responsiveness while reading and layout out the graph
This restructures layoutmore so that it can take a time limit and do
limited amounts of graph layout and graph optimization, and return 1
if it exceeded the time limit before finishing everything it could do.
Also getcommitlines reads at most half a megabyte each time, to limit
the time it spends parsing the commits to about a tenth of a second.
Also got rid of the unused ncmupdate variable while I was at it.
Use the href() function instead of string concatenation to generate
most URLs to our own CGI.
This is a work in progress, not everything has been converted yet.
Signed-off-by: Martin Waitz <tali@admingilde.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
gitweb: provide function to format the URL for an action link.
Provide a new function which can be used to generate an URL for the CGI.
This makes it possible to consolidate the URL generation in order to make
it easier to change the encoding of actions into URLs.
Signed-off-by: Martin Waitz <tali@admingilde.org> Signed-off-by: Junio C Hamano <junkio@cox.net>
* maint:
finish_connect(): thinkofix
git-mv: succeed even if source is a prefix of destination
Solaris does not support C99 format strings before version 10
All but one callers have ignore the return value from this
function, but the only caller, builtin-tar-tree.c::remote_tar(),
assumed it returns non-zero on failure and zero on success. The
implementation however was returning either the waited pid
(which must be the same as its input) or -1 (an error).
Fix this thinko, while getting rid of an assignment of return
value from waitpid() into a variable of type int.
On Solaris nanosleep(2) is not available in libc; you need to
link with -lrt to get it.
The purpose of the loop is to wait until the next filesystem
timestamp granularity, and the code uses subsecond sleep in the
hope that it can shorten the delay to 0.5 seconds on average
instead of a full second. It is probably not worth depending on
an extra library for this.
We might want to yank out the whole "racy-git avoidance is
costly later at runtime, so let's delay writing the index out"
codepath later, but that is a separate issue and needs some
testing on large trees to figure it out. After playing with the
kernel tree, I have a feeling that the whole thing may not be
worth it.