OpenSSH has deprecated its use of scp/rcp protocol in favor of SFTP,
which the embedded Apache mina-sshd in widely-deployed Gerrit
versions does not yet support. The default officially changed in
OpenSSH 9.0 (some distributions, such as Fedora and CentOS, switched
their default behavior to this as early as OpenSSH 8.7 or 8.8),
leading to a ``subsystem request failed on channel 0`` error during
commit-msg hook retrieval. Now git-review will attempt to detect
whether scp's -O option is available to force use of the legacy
scp/rcp protocol, and apply it if so.
Change-Id: Ib64c03c3e12a3a8390e38f6ca9393db3b3c2a9e3
Since mock is a backport of the standard library, and we no longer
support Python 2, we can use the standard library, and drop one external
dependency.
Change-Id: I798c85f1581f4562908c10cd1b58134cdcb40281
The --preserve-merges (-p) option was replaced by --rebase-merges (-r).
This fixes the following error when using git version 2.34.0:
Errors running git rebase -p -i remotes/gerrit/stable/xena
fatal: --preserve-merges was replaced by --rebase-merges
In order to keep compatibility with git < 2.18.0 we detect the git
version and use the old --preserve-merges flag when the version is older
than 2.18.0.
Co-Authored-By: Clark Boylan <clark.boylan@gmail.com>
Change-Id: I04de3d0f20aa6bafcf746b7706d61dd9b9af296c
When checking for unstaged or uncommitted changes to avoid the test
rebase (which could cause data loss for users of git.autostash),
it's still fine if there are unstaged or uncommitted changes in
submodules since those won't be rebased. Have the git diff
invocations explicitly ignore submodules, and also add regression
tests which demonstrate it's working.
This fixes a regression originally introduced by change
Iabb8387c9db59a7d02ebfd43b688e7bb93d3159f.
Change-Id: I20d602e86537b573ac1f9788221215047a594f83
Previously, git-review would assume that the Git repository's hook
directory is .git/hooks, relative to the root of the checkout. This
assumption breaks if the user has set the core.hooksPath option on the
repository (or, for that matter, in ~/.gitconfig or /etc/gitconfig).
core.hooksPath can either be set to an absolute path, in which case it
is to be interpreted as-is, or to a relative path, in which case it
should be interpreted as relative to the root of the checkout.
Introduce a new convenience function to suss out the correct path, and
use it in places where the reference to .git/hooks was previously
hard-coded.
Reference:
https: //git-scm.com/docs/git-config#Documentation/git-config.txt-corehooksPath
Depends-on: I0f0f44e57a100420d8e6d2eaec7dbb5d77b654af
Change-Id: Id8a3ac464ff75e6d8207f198089f018cc790eca5
There is a long standing issue with C Git pushing to Gerrit and Jgit
where the occasional push will fail because the negotiated packs are
missing a tree object. This happens very occasionally but when it does
it would be nice to be able to point users at an easy workaround.
Pushing with --no-thin is that workaround.
Note that --no-thin is much less efficient so shouldn't be used by
default.
This old bug, https://bugs.launchpad.net/git-review/+bug/1332549, has
details but it seems to affect current C git and Gerrit+Jgit.
Change-Id: Id6ba52a656a14c921acab1b14ef668e6251245da
The tests for pushing with unstaged or uncommitted edits had some
left-over code comments from an earlier revision, which were no
longer accurate but also entirely unnecessary. Remove them.
Change-Id: Icfc98d426fa994cce6f1ec290acfa5c0d55123bf
A 2-minute global test timeout was added to the base test class back
in 2014, to prevent tests from hanging until the job timeout is hit.
This has served us well, but with updates to Gerrit and increasing
complexity of our tests, we often get a slew of test timeouts on
some CI nodes (ad on my workstation for that matter!). Increase this
to 5 minutes, which should still serve the intended purpose while
providing more time for tests to complete on less-performant
systems.
Change-Id: I9673ce3cf18f9e4c96fd7e9f96f5ddce2ef6d957
Modernize our package metadata in the following ways:
* switch from description-file to long_description with the file
attribute, and specify an explicit content type and encoding
* replace the home-page parameter with the newer general url one
* add specific labelled project links for improved navigation from
PyPI's summary sidebar
* add commandline keyword to help folks searching
* use the specific license metadata in addition to the corresponding
trove classifier for it
* make sure wheels when built also incorporate the LICENSE and
AUTHORS files so that we're not distributing them without a copy
of the license text
* stop flagging wheels as "universal" now that git-review no longer
supports Python 2.7
* drop the old Sphinx integration config for PBR now that it's no
longer needed
https://setuptools.readthedocs.io/en/latest/userguide/declarative_config.html
Also update old openstack.org URLs throughout contributor docs and
examples/comments to newer opendev.org counterparts. Remove the old
redundant HACKING.rst file as well as a lingering MANIFEST.in from
the times before PBR was a thing. Replace the CONTRIBUTING.rst with
a shorter one cribbed from bindep. Add the test profile to the one
entry in bindep.txt to make it more apparent that's not a runtime
dependency of git-review. Adjust some old "OpenStack, LLC."
copyrights as indicated by the foundation's "Legal Issues FAQ."
Change-Id: Ie45d4d73ba7b5a860f09cc4f1d849587761d846c
For safety, attempts to push a commit with unstaged or uncommitted
changes in the worktree will be caught and an error reported, rather
than leaving it up to ``git rebase`` to spot them. This addresses a
situation where users enabling "rebase.autostash" would otherwise
experience data loss when the test rebase is subsequently reset.
Change-Id: Iabb8387c9db59a7d02ebfd43b688e7bb93d3159f
Task: #38921
In order to better test detection of unstaged and uncommitted
changes, add base test methods to create changes and not stage or
commit them, ensuring they can result in an unstaged diff or staged
but uncommitted edits. To ensure that these work, use them within
the simple change creation method to perform the actual file edit.
Change-Id: Ib698d0057a404f073490d1683a8eef8d0c143122
More recent OpenSSH/OpenSSL versions no longer want to communicate
with the mina-sshd in Gerrit 2.11. In order to pave the way for
testing on distros new enough to supply Python 3.9 packages, upgrade
the version of Gerrit used in the functional testsuite.
Because 2.12 and later moved user SSH keys into Git, we can no
longer rely on the same gsql bootstrapping method to inject them
into the golden site. However, if we init the server in --dev mode,
we can use the default admin/secret test administrator account
instead of creating a new administrator account ourselves (we just
need to post an SSH public key for it to the REST API at runtime).
Also stop precreating host keys and running a reindex, partially
reverting I1c5db09c53a56ffd856da24decf29566e86f9d87, which we did to
keep 2.11 working on some older (but then newer) platforms. Keep the
reindexing in the golden site though, as this remains necessary and
likely also provides some performance benefits.
Add/improve some code comments around relevant bits, so further
updates or rewrites will be less of a struggle for maintainers.
Change-Id: I1245519d912a385c4b9bd09da1728d9b7933802b
In the years since our test framework was created, the
create-project CLI syntax has changed and the --name parameter is
deprecated in favor of using a positional argument. Switching to
this newer syntax becomes necessary for newer Gerrit versions which
no longer recognize the --name option.
Change-Id: I342de827d6a7d7af418f4b1eac54f5dce7c672ed
This adds a setting for choosing which of {name, email, username} to use
as "author" when constructing the name for the branch to where a change
is downloaded.
The rationale is that sometimes the given "name" is just long and
unwieldy (when displaying the branch in the shell prompt, for example),
and one may be already used to the "username".
Change-Id: Ieed465f69ed0c0864979a92f609bd8f58cd8e883
git_credentials() was converting the stdin argument to bytes. However
according to the Python3 documentation subprocess.communicate() expects
a string when universal_newlines=True is passed to Popen.
The symptom was that git review would fail in p.communicate(stdin) on
line 156 with the message "'bytes' object has no attribute 'encode'".
This was observed with Python version 3.6.9 when running git-review
against a gerrit repo over https, where the repo requires username and
password to authenticate.
Change-Id: I0c0314c3f7b0eb631e72e4ac187a9d443a2bc82b
In our setup branch names are not designed to contain topics and are in
certain cases also dictated by other tools. To avoid having git-review
constantly overwrite the topics it is useful to set --no-topic to be the
default state when running git-review.
Change-Id: I1e77e062d73d47f1cceeb34b3418c074d06c9005
Officially removes support for py27, which was already broken for
3+ months (unmaintained test tools).
Change-Id: I38c966d4e1680592f5cd94695051710695d41f71
Related: https://github.com/testing-cabal/subunit/pull/32
As of gerrit 2.15 and later, draft workflow is replaced with
work-in-progress and private workflow. See this CL: [1] and this
issue upstream: [2].
Even though support for draft worklfow was removed, the drafts magic
draft option was preserved, and is mapped to creation of the private
change when pushed first time, or creation of change edit on subsequent
pushes. These behaviour was alaways controvesial, but was kept in place
because 2 major Gerrit clients: repo and git-review were still
referencing drafts magic branch. In upcoming gerrit releases the support
for drafts magic branch option is discontinued, and thus removed from
both repo and git-review: [3].
[1] https://gerrit-review.googlesource.com/c/gerrit/+/97230
[2] https://crbug.com/gerrit/6880
[3] https://gerrit-review.googlesource.com/c/gerrit/+/238898
Sem-Ver: api-break
Change-Id: I08a590d42e1ebaa230da960cd192c0b1df528332
Fixes an issue we're seeing in the python3-based gate plus some other
random things.
Change-Id: I417c0a7669090ee3419c406024f6f3e3289b4c4b
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
When the git configuration value push.followTags is set and a repo has
tags, git-review pushes are rejected:
! [remote rejected] TAG -> TAG
(cannot combine normal pushes and magic pushes)
This change ensures that git-review never pushes with followTags.
Change-Id: Ifbd13284b16bad1165e73d25b99f17344180d423
If usepushurl is set, then it's likely that the pushurl is pointing
to the code review system while the normal url is pointing to a read
only mirror. If the read-only mirror in question doesn't contain
the gerrit refs (like opendev.org currently) then git fetch origin
won't work properly. However, we already calculate the correct
URL at the top of the function for use in query_reviews, and fetch
will happily work with a full url rather than a named remote.
Update the function to use the remote_url so that usepushurl works
even if the mirror url does not contain refs/changes.
Change-Id: Ib72afe97e65cb1dcaf95e28450dfe6e7d5f88965
Our company-internal Gerrit uses federated logins (Shibboleth/SAML), and
the login names happen to include the at-signs. This needed a patch to
Gerrit ([1], released in 2.13.10) to allow them. Cloning using regular
git tools as well as through the gertty work, but `git-review --list`
complained because it couldn't parse these URLs.
[1] https://gerrit-review.googlesource.com/c/94914/
Change-Id: I6c1c75a585184ee3fb2847c1e6d30802e53f8b4c
When something is not found in git config, the --verbose option logs the
wrong value "None", example:
2018-12-12 Running: git config --get gitreview.branch
2018-12-12 using default: None
2018-12-12 Running: git config --get gitreview.scheme
2018-12-12 using default: None
2018-12-12 Running: git config --get remote.gerrit.pushurl
2018-12-12 using default: None
2018-12-12 Running: git config --get remote.gerrit.url
2018-12-12 result: ssh://marc@review.openstack.org:29418/openstack-infra/git-review
From a --verbose user perspective this is plain wrong for (at least) all the
options defined in the DEFAULTS list.
Change --verbose message to look like this instead:
2018-12-12 Running: git config --get gitreview.branch
2018-12-12 Config['branch'] = master
2018-12-12 Running: git config --get gitreview.scheme
2018-12-12 Config['scheme'] = ssh
2018-12-12 Running: git config --get remote.gerrit.pushurl
2018-12-12 Running: git config --get remote.gerrit.url
2018-12-12 ... remote.gerrit.url = ssh://marc@review.openstack.org:29418/openstack-infra/git-review
If git_config_get_value('new_option',... ) is ever invoked in the future
with a not None, default="fubar" parameter then --verbose will print these
lines:
2018-12-19 Running: git config --get gitreview.new_option
2018-12-19 ... nothing in git config, returning func parameter: fubar
This logging issue is especially misleading considering the many levels of
defaults and fallbacks: git config x3; .gitreview; DEFAULTS list,
git_config_get_value(default=...) parameter, options parser logic, etc.
Change-Id: I6cee46e88b90b8f11689be3875d64ec5e577f12f
Newer versions of git quote branch names in their output; at least git
2.16.3 does. This causes all test_uploads_with_nondefault_rebase tests to
fail like this:
testtools.matchers._impl.MismatchError:
'Branch test_branch set up to track remote branch maint from origin.'
not in
u"Switched to a new branch 'test_branch'\n\
Branch 'test_branch' set up to track remote branch 'maint' from 'origin'."
Add a testtools.matchers.MatchesRegex() to support both styles.
Change-Id: I9f1417c53de2f7d638e845f553df3bd426a4c750
From: https://www.openssh.com/txt/release-7.8 change log:
* ssh-keygen(1): write OpenSSH format private keys by default
instead of using OpenSSL's PEM format. The OpenSSH format,
supported in OpenSSH releases since 2014 and described in the
PROTOCOL.key file in the source distribution, offers substantially
better protection against offline password guessing and supports
key comments in private keys. If necessary, it is possible to write
old PEM-style keys by adding "-m PEM" to ssh-keygen's arguments
when generating or updating a key.
This fixes all tests failing with this error:
[2018-12-04 17:46:24,130] WARN org.apache.sshd.common.keyprovider.FileKeyPairProvider :
Unable to read key /home/mherber2/pip/src/git-review/.gerrit/site-5123/etc/ssh_host_rsa_key
java.io.IOException: unrecognised object: OPENSSH PRIVATE KEY
at org.bouncycastle.openssl.PEMParser.readObject(Unknown Source)
at org.apache.sshd.common.keyprovider.FileKeyPairProvider.doLoadKey(FileKeyPairProvider.java:124)
Change-Id: Id10f9b5be928f2ba57847fa32814e9db979375f5
Python2 has default encoding as ascii which means
that is likely that some print() commands would
fails with UnicodeEncodeError.
This hack changes default encoding in order to
avoid such errors.
Change-Id: I4e21e6e32d4bb815693b7d6ce35efb6a5cca2fc2
Refactoring to improve code style as suggested by pep8.
- "rc" and "status" variables cannot be None beacause the process is
terminated after the "run_command_status"
- It's better to ignore an empty "username"
Change-Id: I8a34f4ab9e05d91786ce6d62e4db96811787b633
Unnecessary dynamic GitReviewException.EXIT_CODE attribute
lookup refactored into static to benefit static analyzers.
Default process exit code changed from 127 (command not found error)
to 1 (general error)
Change-Id: I1fcb583a740bf32c4427a587e208d099712a7bc4
