Similar to Bash's default filename completion, our git-aware filename
completion stops at directory boundaries, i.e. it doesn't offer the
full 'path/to/file' at first, but only 'path/'. To achieve that the
completion script runs 'git ls-files' with specific command line
options to get the list of relevant paths under the current directory,
and then processes each path to strip all but the base directory or
filename (see __git_index_files()).
To offer only modified and untracked files for 'git add' the
completion script runs 'git ls-files --exclude-standard --others
--modified'. This command lists all non-ignored files in untracked
directories, which leads to a noticeable delay caused by the
processing mentioned above if there are a lot of such files
(__git_index_files() specifies '--exclude-standard' internally):
$ mkdir untracked-dir
$ for i in {1..10000} ; do >untracked-dir/$i ; done
$ time __git_index_files "--others --modified"
untracked-dir
real 0m0.537s
user 0m0.452s
sys 0m0.160s
Eliminate this delay by additionally passing the '--directory
--no-empty-directory' options to 'git ls-files' to show only the
directory name of non-empty untracked directories instead their whole
content:
$ time __git_index_files "--others --modified --directory --no-empty-directory"
untracked-dir
real 0m0.029s
user 0m0.020s
sys 0m0.004s
Filename completion for 'git clean' suffers from the same delay, as it
offers untracked files, too. The fix could be the same, but since it
actually makes sense to 'git clean' empty directories, in this case we
only pass the '--directory' option to 'git ls-files'.
Reported-by: Isaac Levy <ilevy@google.com>
Signed-off-by: SZEDER Gábor <szeder@ira.uka.de>
Reviewed-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Instead of repeating the text to record as the commit log message
and string we expect to see in "log" output, use the same variable
to avoid them going out of sync.
Use different names for test files in different directories to
improve our chance to catch future breakages that makes "-C <dir>"
go to a place that is different from what was specified.
Signed-off-by: Nazri Ramliy <ayiehere@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Most of git's traversals are robust against minor breakages
in commit data. For example, "git log" will still output an
entry for a commit that has a broken encoding or missing
author, and will not abort the whole operation.
Shortlog, on the other hand, will die as soon as it sees a
commit without an author, meaning that a repository with
a broken commit cannot get any shortlog output at all.
Let's downgrade this fatal error to a warning, and continue
the operation.
We simply ignore the commit and do not count it in the total
(since we do not have any author under which to file it).
Alternatively, we could output some kind of "<empty>" record
to collect these bogus commits. It is probably not worth it,
though; we have already warned to stderr, so the user is
aware that such bogosities exist, and any placeholder we
came up with would either be syntactically invalid, or would
potentially conflict with real data.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A clone will always create a transport struct, whether we
are cloning locally or using an actual protocol. In the
local case, we only use the transport to get the list of
refs, and then transfer the objects out-of-band.
However, there are many options that we do not bother
setting up in the local case. For the most part, these are
noops, because they only affect the object-fetching stage
(e.g., the --depth option). However, some options do have a
visible impact. For example, giving the path to upload-pack
via "-u" does not currently work for a local clone, even
though we need upload-pack to get the ref list.
We can just drop the conditional entirely and set these
options for both local and non-local clones. Rather than
keep track of which options impact the object versus the ref
fetching stage, we can simply let the noops be noops (and
the cost of setting the options in the first place is not
high).
The one exception is that we also check that the transport
provides both a "get_refs_list" and a "fetch" method. We
will now be checking the former for both cases (which is
good, since a transport that cannot fetch refs would not
work for a local clone), and we tweak the conditional to
check for a "fetch" only when we are non-local.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When stderr does not point to a tty, we typically suppress
"we are now in this phase" progress reporting (e.g., we ask
the server not to send us "counting objects" and the like).
The new "checking connectivity" message is in the same vein,
and should be suppressed. Since clone relies on the
transport code to make the decision, we can simply sneak a
peek at the "progress" field of the transport struct. That
properly takes into account both the verbosity and progress
options we were given, as well as the result of isatty().
Note that we do not set up that progress flag for a local
clone, as we do not fetch using the transport at all. That's
acceptable here, though, because we also do not perform a
connectivity check in that case.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Putting messages like "Cloning into.." and "done" on stdout
is un-Unix and uselessly clutters the stdout channel. Send
them to stderr.
We have to tweak two tests to accommodate this:
1. t5601 checks for doubled output due to forking, and
doesn't actually care where the output goes; adjust it
to check stderr.
2. t5702 is trying to test whether progress output was
sent to stderr, but naively does so by checking
whether stderr produced any output. Instead, have it
look for "%", a token found in progress output but not
elsewhere (and which lets us avoid hard-coding the
progress text in the test).
This should not regress any scripts that try to parse the
current output, as the output is already internationalized
and therefore unstable.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some people still use rather old versions of bash, which cannot grok
some constructs like 'printf -v varname' the prompt and completion
code started to use recently.
* bc/completion-for-bash-3.0:
contrib/git-prompt.sh: handle missing 'printf -v' more gracefully
t9902-completion.sh: old Bash still does not support array+=('') notation
git-completion.bash: use correct Bash/Zsh array length syntax
Fixes a minor bug in "git rebase -i" (there could be others, as the
root cause is pretty generic) where the code feeds a random, data
dependeant string to 'echo' and expects it to come out literally.
* mm/no-shell-escape-in-die-message:
die_with_status: use "printf '%s\n'", not "echo"
Output from "git log --full-diff -- <pathspec>" looked strange,
because comparison was done with the previous ancestor that touched
the specified <pathspec>, causing the patches for paths outside the
pathspec to show more than the single commit has changed.
* tr/log-full-diff-keep-true-parents:
log: use true parents for diff when walking reflogs
log: use true parents for diff even when rewriting
The auto-tag-following code in "git fetch" tries to reuse the same
transport twice when the serving end does not cooperate and does
not give tags that point to commits that are asked for as part of
the primary transfer. Unfortunately, Git-aware transport helper
interface is not designed to be used more than once, hence this
does not work over smart-http transfer.
* jc/transport-do-not-use-connect-twice-in-fetch:
builtin/fetch.c: Fix a sparse warning
fetch: work around "transport-take-over" hack
fetch: refactor code that fetches leftover tags
fetch: refactor code that prepares a transport
fetch: rename file-scope global "transport" to "gtransport"
t5802: add test for connect helper
Send a large request to read(2)/write(2) as a smaller but still
reasonably large chunks, which would improve the latency when the
operation needs to be killed and incidentally works around broken
64-bit systems that cannot take a 2GB write or read in one go.
* sp/clip-read-write-to-8mb:
Revert "compat/clipped-write.c: large write(2) fails on Mac OS X/XNU"
xread, xwrite: limit size of IO to 8MB
While normalizing a URL, we forgot that the buffer that holds it
could be relocated when it grows, which was a brown-paper-bag bug
that can lead to a crash introduced on 'master' post 1.8.4 release.
* jc/url-match:
urlmatch.c: recompute pointer after append_normalized_escapes
"git cvsserver" computed the permission mode bits incorrectly for
executable files.
* jc/cvsserver-perm-bit-fix:
cvsserver: pick up the right mode bits
When send-email comes up with an error message to die with upon
failure to start an SSL session, it tried to read the error string
from a wrong place.
* bc/send-email-ssl-die-message-fix:
send-email: don't call methods on undefined values
"git-svn" used with SVN 1.8.0 when talking over https:// connection
dumped core due to a bug in the serf library that SVN uses. Work
it around on our side, even though the SVN side is being fixed.
* uh/git-svn-serf-fix:
git-svn: fix termination issues for remote svn connections
* fc/contrib-bzr-hg-fixes:
contrib/remote-helpers: quote variable references in redirection targets
contrib/remote-helpers: style updates for test scripts
remote-hg: use notes to keep track of Hg revisions
remote-helpers: cleanup more global variables
remote-helpers: trivial style fixes
remote-hg: improve basic test
remote-hg: add missing &&s in the test
remote-hg: fix test
remote-bzr: make bzr branches configurable per-repo
remote-bzr: fix export of utf-8 authors
The implementation of "add -i" has a crippling code to work around
ActiveState Perl limitation but it by mistake also triggered on Git
for Windows where MSYS perl is used.
* js/add-i-mingw:
add--interactive: fix external command invocation on Windows
A packfile that stores the same object more than once is broken and
will be rejected by "git index-pack" that is run when receiving data
over the wire.
* jk/duplicate-objects-in-packs:
t5308: check that index-pack --strict detects duplicate objects
test index-pack on packs with recoverable delta cycles
add tests for indexing packs with delta cycles
sha1-lookup: handle duplicate keys with GIT_USE_LOOKUP
test-sha1: add a binary output mode
We made sure that we notice the user-supplied GIT_DIR is actually a
gitfile, but did not do the same when the default ".git" is a gitfile.
* nd/git-dir-pointing-at-gitfile:
Make setup_git_env() resolve .git file when $GIT_DIR is not specified
If a build-time fallback is set to "cat" instead of "less", we
should apply the same "no subprocess or pipe" optimization as we
apply to user-supplied GIT_PAGER=cat.
* jk/pager-bypass-cat-for-default-pager:
pager: turn on "cat" optimization for DEFAULT_PAGER
Modernize tests.
* fc/rev-parse-test-updates:
rev-parse test: use standard test functions for setup
rev-parse test: use test_cmp instead of "test" builtin
rev-parse test: use test_must_fail, not "if <command>; then false; fi"
rev-parse test: modernize quoting and whitespace
When --shallow-file is added to the command line, it has to be
before the subcommand name, the first argument won't be the command
name any more. Stop assuming that and keep track of the command name
explicitly.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Unlike other git commands which work correctly at the top-level or in a
subdirectory, git-contacts fails when invoked in a subdirectory. This is
because it invokes git-blame with pathnames relative to the top-level,
but git-blame interprets the pathnames as relative to the current
directory. Fix this.
Signed-off-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
By doing this, clients of upload-pack can now reliably tell what ref
a symbolic ref points at; the updated test in t5505 used to expect
failure due to the ambiguity and made sure we give diagnostics, but
we no longer need to be so pessimistic. Make sure we correctly learn
which branch HEAD points at from the other side instead.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With the same mechanism as used to tell where "HEAD" points at to
the other end, we can tell the target of other symbolic refs as
well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
One long-standing flaw in the pack transfer protocol was that there
was no way to tell the other end which branch "HEAD" points at.
With a capability "symref=HEAD:refs/heads/master", let the sender to
tell the receiver what symbolic ref points at what ref.
This capability can be repeated more than once to represent symbolic
refs other than HEAD, such as "refs/remotes/origin/HEAD").
Add an infrastructure to collect symbolic refs, format them as extra
capabilities and put it on the wire. For now, just send information
on the "HEAD" and nothing else.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The callee does not use cb_data, and the caller is an intermediate
function in a callchain that later wants to use the cb_data for its
own use. Clarify the code by breaking the dataflow explicitly by
not passing cb_data down to mark_our_ref().
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When two or more branches point at the same commit and HEAD is
pointing at one of them, without the symref extension, there is no
way to remotely tell which one of these branches HEAD points at.
The test in question attempts to make sure that this situation is
diagnosed and results in a failure.
However, even if there _were_ a way to reliably tell which branch
the HEAD points at, "set-head --auto" would fail if there is no
remote tracking branch. Make sure that this test does not fail
for that "wrong" reason.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
During the review process of the previous commit (repack: rewrite the
shell script in C), Johannes Sixt proposed to retain any exit codes from
the sub-process, which makes it probably more obvious in case of failure.
As the commit before should behave as close to the original shell
script, the proposed change is put in this extra commit.
The infrastructure however was already setup in the previous commit.
(Having a local 'ret' variable)
Signed-off-by: Stefan Beller <stefanbeller@googlemail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The motivation of this patch is to get closer to a goal of being
able to have a core subset of git functionality built in to git.
That would mean
* people on Windows could get a copy of at least the core parts
of Git without having to install a Unix-style shell
* people using git in on servers with chrooted environments
do not need to worry about standard tools lacking for shell
scripts.
This patch is meant to be mostly a literal translation of the
git-repack script; the intent is that later patches would start using
more library facilities, but this patch is meant to be as close to a
no-op as possible so it doesn't do that kind of thing.
Signed-off-by: Stefan Beller <stefanbeller@googlemail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove now disused remote-helpers framework for helpers written in
Python.
* jk/remove-remote-helpers-in-python:
git_remote_helpers: remove little used Python library
When we converted many documents that were traditionally text-only
to be formatted to AsciiDoc, we did not update links that point at
them to refer to the formatted HTML files.
* ss/doclinks:
Documentation: make AsciiDoc links always point to HTML files
We liberally use "committish" and "commit-ish" (and "treeish" and
"tree-ish"); as these are non-words, let's unify these terms to
their dashed form. More importantly, clarify the documentation on
object peeling using these terms.
* rh/ishes-doc:
glossary: fix and clarify the definition of 'ref'
revisions.txt: fix and clarify <rev>^{<type>}
glossary: more precise definition of tree-ish (a.k.a. treeish)
use 'commit-ish' instead of 'committish'
use 'tree-ish' instead of 'treeish'
glossary: define commit-ish (a.k.a. committish)
glossary: mention 'treeish' as an alternative to 'tree-ish'
When the user types "git diff" outside a working tree, thinking he
is inside one, the current error message that is a single-liner
"usage: git diff --no-index <path> <path>" may not be sufficient to
make him realize the mistake. Add "Not a git repository" to the
error message when we fell into the "--no-index" mode without an
explicit command line option to instruct us to do so.
* dw/diff-no-index-doc:
diff --no-index: describe in a separate paragraph
diff --no-index: clarify operation when not inside a repository
Update the user's manual to more recent versions of Git.
* ta/user-manual:
"git prune" is safe
Remove irrelevant reference from "Tying it all together"
Remove unnecessary historical note from "Object storage format"
Improve section "Merging multiple trees"
Improve section "Manipulating branches"
Simplify "How to make a commit"
Fix some typos and improve wording
Use "git merge" instead of "git pull ."
Use current output for "git repack"
Use current "detached HEAD" message
Call it "Git User Manual" and remove reference to very old Git version
