This addresses the issue where Git for Windows asks the user for a
password, no credential helper is available, and then Git fails to pick
up non-ASCII characters from the Git GUI helper.
This can be verified e.g. via
echo host=http://abc.com |
git -c credential.helper= credential fill
and then pasting some umlauts.
The underlying reason is that Git for Windows tries to communicate using
the UTF-8 encoding no matter what the actual current code page is. So
let's indulge Git for Windows and do use that encoding.
This fixes https://github.com/git-for-windows/git/issues/2215
Signed-off-by: Luke Bonanomi <lbonanomi@gmail.com>
Helped-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Pratyush Yadav <me@yadavpratyush.com>
Fixes an error popup in blame because of a missing closing bracket.
* py/blame-status-error:
git-gui: fix error popup when doing blame -> "Show History Context"
Several tests wanted to verify that files were actually modified by a
merge, which it would do by checking that the mtime was updated. In
order to avoid problems with the merge completing so fast that the mtime
at the beginning and end of the operation was the same, these tests
would first set the mtime of a file to something "old". This "old"
value was usually determined as current system clock minus one second,
truncated to the nearest integer. Unfortunately, it appears the system
clock and filesystem clock are different and comparing across the two
runs into race problems resulting in flaky tests.
From https://stackoverflow.com/questions/14392975/timestamp-accuracy-on-ext4-sub-millsecond:
date will call the gettimeofday system call which will always return
the most accurate time available based on the cached kernel time,
adjusted by the CPU cycle time if available to give nanosecond
resolution. The timestamps stored in the file system however, are
only based on the cached kernel time. ie The time calculated at the
last timer interrupt.
and from https://apenwarr.ca/log/20181113:
Does mtime get set to >= the current time?
No, this depends on clock granularity. For example, gettimeofday()
can return times in microseconds on my system, but ext4 rounds
timestamps down to the previous ~10ms (but not exactly 10ms)
increment, with the surprising result that a newly-created file is
almost always created in the past:
$ python -c "
import os, time
t0 = time.time()
open('testfile', 'w').close()
print os.stat('testfile').st_mtime - t0
"
-0.00234484672546
So, instead of trying to compare across what are effectively two
different clocks, just avoid using the system clock. Any new updates to
files have to give an mtime at least as big as what is already in the
file, so we could define "old" as one second before the mtime found in
the file before the merge starts. But, to avoid problems with leap
seconds, ntp updates, filesystems that only provide two second
resolution, and other such weirdness, let's just pick an hour before the
mtime found in the file before the merge starts.
Also, clarify in one test where we check the mtime of different files
that it really was intentional. I totally forgot the reasons for that
and assumed it was a bug when asked.
Reported-by: SZEDER Gábor <szeder.dev@gmail.com>
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Band-aid fixes for two fallouts from switching the default "rebase"
backend.
* en/rebase-backend:
git-rebase.txt: highlight backend differences with commit rewording
sequencer: clear state upon dropping a become-empty commit
i18n: unmark a message in rebase.c
The credential protocol can't handle values with newlines. We already
detect and block any such URLs from being used with credential helpers,
but let's also add an fsck check to detect and block gitmodules files
with such URLs. That will let us notice the problem earlier when
transfer.fsckObjects is turned on. And in particular it will prevent bad
objects from spreading, which may protect downstream users running older
versions of Git.
We'll file this under the existing gitmodulesUrl flag, which covers URLs
with option injection. There's really no need to distinguish the exact
flaw in the URL in this context. Likewise, I've expanded the description
of t7416 to cover all types of bogus URLs.
The credential protocol can't represent newlines in values, but URLs can
embed percent-encoded newlines in various components. A previous commit
taught the low-level writing routines to die() when encountering this,
but we can be a little friendlier to the user by detecting them earlier
and handling them gracefully.
This patch teaches credential_from_url() to notice such components,
issue a warning, and blank the credential (which will generally result
in prompting the user for a username and password). We blank the whole
credential in this case. Another option would be to blank only the
invalid component. However, we're probably better off not feeding a
partially-parsed URL result to a credential helper. We don't know how a
given helper would handle it, so we're better off to err on the side of
matching nothing rather than something unexpected.
The die() call in credential_write() is _probably_ impossible to reach
after this patch. Values should end up in credential structs only by URL
parsing (which is covered here), or by reading credential protocol input
(which by definition cannot read a newline into a value). But we should
definitely keep the low-level check, as it's our final and most accurate
line of defense against protocol injection attacks. Arguably it could
become a BUG(), but it probably doesn't matter much either way.
Note that the public interface of credential_from_url() grows a little
more than we need here. We'll use the extra flexibility in a future
patch to help fsck catch these cases.
The credential tests have a "check" function which feeds some input to
git-credential and checks the stdout and stderr. We look for exact
matches in the output. For stdout, this makes sense; the output is
the credential protocol. But for stderr, we may be showing various
diagnostic messages, or the prompts fed to the askpass program, which
could be translated. Let's mark them as such.
The credential protocol that we use to speak to helpers can't represent
values with newlines in them. This was an intentional design choice to
keep the protocol simple, since none of the values we pass should
generally have newlines.
However, if we _do_ encounter a newline in a value, we blindly transmit
it in credential_write(). Such values may break the protocol syntax, or
worse, inject new valid lines into the protocol stream.
The most likely way for a newline to end up in a credential struct is by
decoding a URL with a percent-encoded newline. However, since the bug
occurs at the moment we write the value to the protocol, we'll catch it
there. That should leave no possibility of accidentally missing a code
path that can trigger the problem.
At this level of the code we have little choice but to die(). However,
since we'd not ever expect to see this case outside of a malicious URL,
that's an acceptable outcome.
Reported-by: Felix Wilhelm <fwilhelm@google.com>
As noted by Junio:
Back when "git am" was written, it was not considered a bug that the
"git am --resolved" option did not offer the user a chance to update
the log message to match the adjustment of the code the user made,
but honestly, I'd have to say that it is a bug in "git am" in that
over time it wasn't adjusted to the new world order where we
encourage users to describe what they did when the automation
hiccuped by opening an editor. These days, even when automation
worked well (e.g. a clean auto-merge with "git merge"), we open an
editor. The world has changed, and so should the expectations.
Junio also suggested providing a workaround such as allowing --no-edit
together with git rebase --continue, but that should probably be done in
a patch after the git-2.26.0 release. For now, just document the known
difference in the Behavioral Differences section.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In commit e98c4269c8 ("rebase (interactive-backend): fix handling of
commits that become empty", 2020-02-15), the merge backend was changed
to drop commits that did not start empty but became so after being
applied (because their changes were a subset of what was already
upstream). This new code path did not need to go through the process of
creating a commit, since we were dropping the commit instead.
Unfortunately, this also means we bypassed the clearing of the
CHERRY_PICK_HEAD and MERGE_MSG files, which if there were no further
commits to cherry-pick would mean that the rebase would end but assume
there was still an operation in progress. Ensure that we clear such
state files when we decide to drop the commit.
Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Commit v2.25.0-4-ge98c4269c8 (rebase (interactive-backend): fix handling
of commits that become empty, 2020-02-15) marked "{drop,keep,ask}" for
translation, but this message should not be changed.
Signed-off-by: Jiang Xin <worldhello.net@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Both "git ls-remote -h" and "git grep -h" give short usage help,
like any other Git subcommand, but it is not unreasonable to expect
that the former would behave the same as "git ls-remote --head"
(there is no other sensible behaviour for the latter). The
documentation has been updated in an attempt to clarify this.
* jc/doc-single-h-is-for-help:
Documentation: clarify that `-h` alone stands for `help`
* 'master' of github.com:git/git: (27 commits)
Git 2.26-rc1
remote-curl: show progress for fetches over dumb HTTP
show_one_mergetag: print non-parent in hex form.
config.mak.dev: re-enable -Wformat-zero-length
rebase-interactive.c: silence format-zero-length warnings
mingw: workaround for hangs when sending STDIN
t6020: new test with interleaved lexicographic ordering of directories
t6022, t6046: test expected behavior instead of testing a proxy for it
t3035: prefer test_must_fail to bash negation for git commands
t6020, t6022, t6035: update merge tests to use test helper functions
t602[1236], t6034: modernize test formatting
merge-recursive: apply collision handling unification to recursive case
completion: add diff --color-moved[-ws]
t1050: replace test -f with test_path_is_file
am: support --show-current-patch=diff to retrieve .git/rebase-apply/patch
am: support --show-current-patch=raw as a synonym for--show-current-patch
am: convert "resume" variable to a struct
parse-options: convert "command mode" to a flag
parse-options: add testcases for OPT_CMDMODE()
stash push: support the --pathspec-from-file option
...
"git fetch" over HTTP walker protocol did not show any progress
output. We inherently do not know how much work remains, but still
we can show something not to bore users.
* rs/show-progress-in-dumb-http-fetch:
remote-curl: show progress for fetches over dumb HTTP
"git show" and others gave an object name in raw format in its
error output, which has been corrected to give it in hex.
* hd/show-one-mergetag-fix:
show_one_mergetag: print non-parent in hex form.
Recently we inadvertently added a few instances of using 0-width
format string to functions that we mark as printf-like without any
developers noticing. The root cause was that the compiler warning
that is triggered by this is almost always useless and we disabled
the warning in our developer builds, but not for general public.
The new instances have been corrected, and the warning has been
resurrected in the developer builds.
* rt/format-zero-length-fix:
config.mak.dev: re-enable -Wformat-zero-length
rebase-interactive.c: silence format-zero-length warnings
Test cleanup.
* en/test-cleanup:
t6020: new test with interleaved lexicographic ordering of directories
t6022, t6046: test expected behavior instead of testing a proxy for it
t3035: prefer test_must_fail to bash negation for git commands
t6020, t6022, t6035: update merge tests to use test helper functions
t602[1236], t6034: modernize test formatting
Handling of conflicting renames in merge-recursive have further
been made consistent with how existing codepaths try to mimic what
is done to add/add conflicts.
* en/merge-path-collision:
merge-recursive: apply collision handling unification to recursive case
"git am --short-current-patch" is a way to show the piece of e-mail
for the stopped step, which is not suitable to directly feed "git
apply" (it is designed to be a good "git am" input). It learned a
new option to show only the patch part.
* pb/am-show-current-patch:
am: support --show-current-patch=diff to retrieve .git/rebase-apply/patch
am: support --show-current-patch=raw as a synonym for--show-current-patch
am: convert "resume" variable to a struct
parse-options: convert "command mode" to a flag
parse-options: add testcases for OPT_CMDMODE()
We grep for "File exists" in stderr of the failing `git sparse-checkout`
to make sure that it failed for the right reason. We expect the string
to show up there since we call `strerror(errno)` in
`unable_to_lock_message()` in lockfile.c.
On the NonStop platform, this fails because the error string is "File
already exists", which doesn't match our grepping.
See 9042140097 ("test-dir-iterator: do not assume errno values",
2019-07-30) for a somewhat similar fix. There, we patched a test helper,
which meant we had access to `errno` and could investigate it better in
the test helper instead of just outputting the numerical value and
evaluating it in the test script. The current situation is different,
since (short of modifying the lockfile machinery, e.g., to be more
verbose) we don't have more than the output from `strerror()` available.
Except we do: We prefix `strerror(errno)` with `_("Unable to create
'%s.lock': ")`. Let's grep for that part instead. It verifies that we
were indeed unable to create the lock file. (If that fails for some
other reason than the file existing, we really really should expect
other tests to fail as well.)
An alternative fix would be to loosen the expression a bit and grep for
"File.* exists" instead. There would be no guarantee that some other
implementation couldn't come up with another error string, That is, that
could be the first move in an endless game of whack-a-mole. Of course,
it could also take us from "99" to "100" percent of the platforms and
we'd never have this problem again. But since we have another way of
addressing this, let's not even try the "loosen it up a bit" strategy.
Reported-by: Randall S. Becker <rsbecker@nexbridge.com>
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Acked-by: Derrick Stolee <dstolee@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
