John Keeping [Thu, 13 Aug 2015 11:14:13 +0000 (12:14 +0100)]
tests: allow shell to be overridden
On some systems (e.g. Solaris), /bin/sh is not a POSIX shell. Git
already provides suitable overrides in its config.mak.uname file and we
provide cgit.conf to allow the user to further change this.
The code for this is taken from Git's t/Makefile, meaning that we now
invoke the tests in the same way that Git does.
John Keeping [Wed, 12 Aug 2015 14:55:28 +0000 (15:55 +0100)]
log: allow users to follow a file
Teach the "log" UI to behave in the same way as "git log --follow", when
given a suitable instruction by the user. The default behaviour remains
to show the log without following renames, but the follow behaviour can
be activated by following a link in the page header.
Follow is not the default because outputting merges in follow mode is
tricky ("git log --follow" will not show merges). We also disable the
graph in follow mode because the commit graph is not simplified so we
end up with frequent gaps in the graph and many lines that do not
connect with any commits we're actually showing.
We also teach the "diff" and "commit" UIs to respect the follow flag on
URLs, causing the single-file version of these UIs to detect renames.
This feature is needed only for commits that rename the path we're
interested in.
For commits before the file has been renamed (i.e. that appear later in
the log list) we change the file path in the links from the log to point
to the old name; this means that links to commits always limit by the
path known to that commit. If we didn't do this we would need to walk
down the log diff'ing every commit whenever we want to show a commit.
The drawback is that the "Log" link in the top bar of such a page links
to the log limited by the old name, so it will only show pre-rename
commits. I consider this a reasonable trade-off since the "Back" button
still works and the log matches the path displayed in the top bar.
Since following renames requires running diff on every commit we
consider, I've added a knob to the configuration file to globally
enable/disable this feature. Note that we may consider a large number
of commits the revision walking machinery no longer performs any path
limitation so we have to examine every commit until we find a page full
of commits that affect the target path or something related to it.
Suggested-by: René Neumann <necoro@necoro.eu> Signed-off-by: John Keeping <john@keeping.me.uk>
John Keeping [Tue, 23 Dec 2014 14:40:44 +0000 (14:40 +0000)]
contrib/hooks: add sample post-receive hook using agefile
One of the most frequent questions on the mailing list relates to the
idle time in the repository list. The answer to this is to use the
"agefile" feature to calculate the time of the last change whenever the
repository receives changes.
Add a sample post-receive hook in a new "contrib" directory so that we
can just point people at the repository in the future.
John Keeping [Sun, 8 Mar 2015 16:32:27 +0000 (16:32 +0000)]
html: avoid using a plain integer as a NULL pointer
Sparse complains about this table because we use the integer zero as the
NULL pointer. Use this as an opportunity to reformat the table so that
it always contains 8 elements per row, making it easier to see which
values are being set and which are not.
John Keeping [Sun, 8 Mar 2015 16:32:24 +0000 (16:32 +0000)]
ui-shared: avoid initializing static variable to zero
Sparse complains that we are using a plain integer as a NULL pointer
here, but in fact we do not have to specify a value for this variable at
all since it has static storage duration and thus will be initialized to
NULL by the compiler.
Lukas Fleischer [Thu, 5 Mar 2015 11:58:12 +0000 (12:58 +0100)]
Drop return value from parse_user()
In commit 936295c (Simplify commit and tag parsing, 2015-03-03), the
commit and tag parsing code was refactored. This broke tag messages in
ui-tag since the line after the tagger header was erroneously skipped.
Rework parse_user() and skip the line manually outside parse_user().
John Keeping [Tue, 3 Mar 2015 19:22:31 +0000 (19:22 +0000)]
cache: use F_SETLK to avoid stale lock files
If CGit is killed while it holds a lock on a cache slot (for example
because it is taking too long to generate a page), the lock file will be
left in place. This prevents any future attempt to use the same slot
since it will fail to exclusively create the lock file.
Since CGit is the only program that should be manipulating lock files,
we can use advisory locking to detect whether another process is
actually using the lock file or if it is now stale.
I have confirmed that this works on Linux by setting a short TTL in a
custom cgitrc and running the following with CGit patched to print a
message to stderr if the fcntl(2) fails:
Lukas Fleischer [Tue, 3 Mar 2015 12:00:07 +0000 (13:00 +0100)]
Simplify commit and tag parsing
* Use skip_prefix to avoid magic numbers in the code.
* Use xcalloc() instead of xmalloc(), followed by manual initialization.
* Split out line splitting.
Christian Hesse [Wed, 11 Feb 2015 08:12:44 +0000 (09:12 +0100)]
shrink cgit.png file size
Ran optipng against cgit.png, which shrank file size by more than eight
percent. The image (including protocol overhead) should fit into a
single network packet now.
Optipng optimizes filters and compression. The actual pixel results are
not altered.
Lukas Fleischer [Mon, 9 Feb 2015 06:25:00 +0000 (07:25 +0100)]
ui-clone.c: Fix path check
The starts_with() check was broken in two ways: For one thing, the
parameters were passed in the wrong order, for another thing,
starts_with() returns 1 if the string starts with the prefix (not 0).
Note that this bug existed since commit 02a545e (Add support for cloning
over http, 2008-08-06) but only pops in in corner cases.
Christian Hesse [Sat, 7 Feb 2015 13:18:28 +0000 (14:18 +0100)]
git: update for v2.3.0
* sort_string_list(): rename to string_list_sort() (upstream commit 3383e199)
* update read_tree_recursive callback to pass strbuf as base (upstream
commit 6a0b0b6d)
Lukas Fleischer [Thu, 29 Jan 2015 11:52:49 +0000 (12:52 +0100)]
Add repo.hide and repo.ignore
These options can be used to hide a repository from the index or
completely ignore a repository, respectively. They are particularly
useful when used in combination with scan-path.
John Keeping [Thu, 15 Jan 2015 22:18:14 +0000 (22:18 +0000)]
tag: reference with "h" instead of "id"
When clicking on "log" from a tag we end up showing the log of whatever
branch we used to reach the tag. If the tag doesn't point onto a branch
then the tagged commit won't appear in this output.
By linking to tags with the head parameter instead of the "id" parameter
the log link will show the log of the tag. This is clearly desirable
when the tag has been reached from the refs UI and changing the
behaviour for tag decorations makes them match branch decorations where
log -> decoration -> log shows the log of the decoration.
Reported-by: Ferry Huberts <mailings@hupie.com> Signed-off-by: John Keeping <john@keeping.me.uk>
Lukas Fleischer [Thu, 15 Jan 2015 18:47:42 +0000 (19:47 +0100)]
Return proper HTTP response when accessing info/
Currently, when a user directly accesses the info command of a
repository, we exit cgit without printing anything to stdout, bringing
up error messages like "502 Bad Gateway" or "An error occurred while
reading CGI reply (no response received)". Instead of bailing out, at
least print the HTTP headers, including a reasonable error message.
John Keeping [Mon, 29 Dec 2014 22:27:55 +0000 (22:27 +0000)]
ui-diff: don't link to single file diff stat
Seeing the diff stat for a single file is pretty useless, so reset the
diff type before generating the links to individual files in the diff
stat so that the links will show a useful diff.
Reported-by: Konstantin Ryabitsev <mricon@kernel.org> Signed-off-by: John Keeping <john@keeping.me.uk>
John Keeping [Sun, 28 Dec 2014 13:10:33 +0000 (13:10 +0000)]
ui-patch: match git-format-patch(1) output
Using (DIFF_FORMAT_DIFFSTAT | DIFF_FORMAT_PATCH) causes Git to emit a
"---" line between the commit message and the body of the patch, which
fixes a regression introduced in commit 455b598 (ui-patch.c: Use
log_tree_commit() to generate diffs, 2013-08-20), prior to which we
inserted the "---" line ourselves.
DIFF_FORMAT_SUMMARY is added so that we match the output of
git-format-patch(1) without the "-p" option.
Lukas Fleischer [Wed, 24 Dec 2014 07:50:11 +0000 (08:50 +0100)]
Use split_ident_line() in parse_user()
Use Git's built-in ident line splitting algorithm instead of
reimplementing it. This does not only simplify the code but also makes
sure that cgit is consistent with Git when it comes to author parsing.
John Keeping [Fri, 1 Aug 2014 21:14:17 +0000 (22:14 +0100)]
Extract clone URL printing to ui-shared.c
This will allow us to reuse the same logic to add clone URL <link/>
elements to the header of all repo-specific pages in order to support
the rel-vcs microformat.
Christian Hesse [Wed, 10 Sep 2014 09:24:07 +0000 (11:24 +0200)]
filter: fix libravatar email-filter https issue
Serving cgit via https and getting avatar via http gives error messages
about untrusted content. This decides whether or not to use https link
by looking at the environment variable HTTPS, which is set in CGI.
John Keeping [Sun, 5 Oct 2014 09:59:04 +0000 (10:59 +0100)]
Change "ss" diff flag to an enum
This will allow us to introduce a new "stat only" diff mode without
needing an explosion of mutually incompatible flags.
The old "ss" query parameter is still accepted in order to avoid
breaking saved links, but we no longer generate any URIs using it;
instead the new "dt" (diff type) parameter is used.
Specifying a nonexistent README file via the readme option is sometimes
useful, e.g. when using scan-path and setting a global default.
Currently, we check whether there is only one option in the readme
option and, if so, we choose that file without checking whether it
exists. As a consequence, all repositories are equipped with an about
link in the aforementioned scenario, even if there is no about file.
Remove the early check for the number of keys and always check whether
the file exists instead.
John Keeping [Sun, 27 Jul 2014 10:56:19 +0000 (11:56 +0100)]
git: update to v2.0.3
This is slightly more involved than just bumping the version number
because it pulls in a change to convert the commit buffer to a slab,
removing the "buffer" field from "struct commit". All sites that access
"commit->buffer" have been changed to use the new functions provided for
this purpose.
Length truncated to 4096. The cache on disk looks truncated as well, so
the bug must me during the process of saving cache. The same is true for
larger patches:
Since the "html" functions use raw write(2) to STDIO_FILENO, we don't
notice problems with most pages, but raw patches write using printf(3).
This is fine if we're outputting straight to stdout since the buffers
are flushed on exit, but we close the cache output before this, so the
cached output ends up being truncated.
Make sure the buffers are flushed when we finish outputting a patch so
that we avoid this.
No other UIs use printf(3) so we do not need to worry about them.
Actually, it's slightly more interesting than this... since we don't set
GIT_FLUSH, Git decides whether or not it will flush stdout after writing
each commit based on whether or not stdout points to a regular file (in
maybe_flush_or_die()).
Which means that when writing directly to the webserver, Git flushes
stdout for us, but when we redirect stdout to the cache it points to a
regular file so Git no longer flushes the output for us.
The patch is still correct, but perhaps the full explanation is
interesting!
Reported-by: Konstantin Ryabitsev <mricon@kernel.org>
Using something like "range" and "qwerty123456", it returns an "Internal
Server Error" and the following in the logs:
> [Tue Jun 10 17:45:32 2014] [error] [client 172.21.1.6] fatal:
> ambiguous argument 'qwerty123456': unknown revision or path not in the
> working tree., referer:
> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/
> [Tue Jun 10 17:45:32 2014] [error] [client 172.21.1.6] Use '--' to
> separate paths from revisions, like this:, referer:
> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/
> [Tue Jun 10 17:45:32 2014] [error] [client 172.21.1.6] 'git <command>
> [<revision>...] -- [<file>...]', referer:
> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/
> [Tue Jun 10 17:45:32 2014] [error] [client 172.21.1.6] Premature end
> of script headers: cgit, referer:
> http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/
The cache will kick in, so if you search for the same string again,
it'll show an empty range, so you have to change the bogus strings each
time.
This is because we just pass the arguments straight to Git's revision
parsing machinery which die()s if it cannot parse an argument, printing
the above to stderr and exiting.
The patch below makes it a bit friendlier by just ignoring unhandled
arguments, but I can't see an easy way to report errors when we can't
parse revision arguments without losing the flexibility of supporting
all of the revision specifiers supported by Git.
Reported-by: Konstantin Ryabitsev <mricon@kernel.org>
Christian Hesse [Thu, 29 May 2014 15:35:46 +0000 (17:35 +0200)]
git: update for git 2.0
prefixcmp() and suffixcmp() have been remove, functionality is now
provided by starts_with() and ends_with(). Retrurn values have been
changed, so instead of just renaming we have to fix logic.
Everything else looks just fine.
Christian Hesse [Thu, 20 Feb 2014 19:48:45 +0000 (20:48 +0100)]
print download link for reference string length == 1
I have a number of repositories that start tagging with just '1' and
count up. Actually references with sting length of one are skipped, this
patch changes that.