The encoding header is semi-invalid has the encoding name should be
"utf-8" and not "utf8", which is not the official name of the encoding
and is not recognized by all tools.
Change-Id: I4db673fe86f643f5e1b3e5135d8b0cb72407253e
Replace URLs for workflow documentation to appropriate parts of the
OpenStack Project Infrastructure Manual.
Change-Id: If393085cacf4d2fa21ae6c49ffa6290e02659588
Gerrit 2.9 adds the '--all-reviewers' option to 'gerrit query', which
allows testing that reviewers have been successfully added to a patch
set in tests for new reviewer functionality.
Change-Id: I17f37c8fd46578b4a8cbfacee4eb8073972b2eee
Need to override both variations of the no_proxy environment variable
since some tools do not check for the other if one is already defined.
While 'no_proxy' appears to be now the most common convention, it is
safer to set both lower and upper case to the same value when running
unit tests.
Change-Id: I536b6f2711ac70e0a2a45676098fb461ab3b8d33
Use the git config command and provided options to have git detect
whether color support should be enabled or disabled. Additionally ensure
that git-review color support can be controlled separately via the
option 'color.review', while still falling back to 'color.ui' if the app
specific option is not defined.
This will ensure that when piping or redirecting the output that colour
output will be disabled unless color.review, or it's fallback color.ui,
is set to always.
Also by utilizing 'git config', the any config options specified on the
command line that change behaviour will be adhered to. e.g.
git -c color.review=never review -l
Will always output without color enabled.
Change-Id: I8f83ed8623da88e87972109af956331704e15d38
Closes-Bug: #1097961
git-review retrieved the remote URL based on the output of git-config.
However, contrary to the previous implementation based on
git-remote-show, git-config does not expand the URL of the given remote
taking into account any "url.<base>.insteadOf" config setting.
This change adds the necessary circuitry to expand URLs.
It also adds a regression test.
Change-Id: I826251aaf85d1b1e8bb2b4562e4b65025a7091b9
Closes-Bug: #1348634
By using the GitReviewException class and subclasses, the raised exceptions
will be caught in main() and properly handled. One Exception can be
converted to a CommandFailed exception, and the other is simple enough
to not need a subclass of GitReviewException.
Change-Id: Idb02d43fd480ab67493c757fb95431602239a55f
git-review gets its configuration from multiple sources (by priority):
* command line options
* git-config
* options: username, rebase
* with local/global/system config
* git-review config files
* options: scheme, host, port, project, defaultbranch/remote/rebase
* locally using .gitreview
* globally using ~/.config/gitreview/git-review.conf
* system using /etc/git-review/git-review.conf
.gitreview file is commonly provided versioned in the git repo, where
other git-review config files and git-config is done by developers. It
implies for developers multiple places to define local/global/system
configuration options but some can only be changed by updating
.gitreview (scheme/port...) which is under versioning.
This change proposes to use as configuration sources (by priority):
* command line options
* git-config for developer local/global/system config
* options: username, rebase and all .gitreview options
* .gitreview local file for repo provided specific config
and deprecates git-review global/system files in order to reduce
configuration sources and allow to overload .gitreview configuration
without updating it.
Change-Id: I24aba0fa089de57d51a79049d9c192f76d576f69
git-review behavior depends on git local, global and system configs.
tests have the same dependencies against them but only manage git local
config which implies the config expected by tests could not be the real
one because of git global and system configs.
The same trouble concerns gitreview global and system config files.
This change allows to run git-review in local mode (only using git and
gitreview local config files) by defining the environment variable
"GITREVIEW_LOCAL_MODE". This mode is used by tests to isolate them from
git and gitreview global and system configs. Mocking can not be used as
functional tests call not git-review code but git-review system command.
Change-Id: I598a284aec9e55f9618ab026c55ef0435e370d69
Closes-Bug: #1393192
Currently, python -m git_review.cmd wraps git_review.cmd.main call to
handle correctly unicode in python2 where git-review calls directly
git_review.cmd.main. This change embeds the wrapper inside main in
order to align git-review and python -m git_review.cmd behaviors.
Change-Id: I390e2d68b851299c3b07292b89dee8198068e6aa
The option -T/--no-topic disables review with topic (mutually
exclusive with -t/--topic).
It allows to submit:
* a change without your current branch name as topic,
* a change update without updating current topic.
Change-Id: I06309b14246cb455d55c6833ab6adb932995edfa
--force-rebase has no effect if rebase is disabled through git-config
or git-review config files.
This change ensures -F/--force-rebase "wins" over rebase disabled and
-F/--force-rebase becomes mutually exclusive with -R/--no-rebase.
Change-Id: Idd985a6a7a5b1cffa9bed996bb1ea5907e7d0844
If git-review is not run in git repo top-level dir, then the target
path used for scp uses both "/" and "\", and scp fails.
Change-Id: I8ffe163f9e29423b1c19bd54241a526a9843561f
Closes-Bug: #1024073
Previous addition of backslashes '\' to escape hypens '-' also escaped
options to the mdoc formatting commands by accident.
Change-Id: I327ccb8a84fb199362e6a5dd7c5f5bff3d5b2bb4
If the last field of the review being listed exceeds the width of the
console, then using whitespace padding will result in sufficient spacing
being added to all other line that will appear as spurious blank lines
if they were otherwise less than the console width.
Instead of:
1234 master <subject>
5678 master <really-long-subject-that-
wraps-to-next-line>
Display:
1234 master <subject>
5678 master <really-long-subject-that-
wraps-to-next-line>
Change-Id: I290f652f803cf3938a1bdb72c20c85969dbb3319
Gerrit 2.8.6.1 bundles an updated mina-sshd which solves the
1-in-256 SSH hash mismatch bug, so should make our tests much more
stable.
Change-Id: Ic607660af8feb5c2dc4d1fca2681265eefdc7e93
Since git-review is tested and works with both Python 2.x and 3.x
interpreters, we want to build py2.py3-none-any wheels rather than
just py2-none-any. Without this change the wheels created are only
retrieved by pip running under a Python 2.x interpreter, while
Python 3.x environments retrieve the tarball instead.
Change-Id: I4d0cd28095b9c0273c11f2a1a4b152ba3b2dca33
The homepage entry displayed on PyPI is sort of useless if it points
to itself. Instead use OpenStack's cgit site for it.
Change-Id: I8a82418234058716b55cd5e23d114ba30d3ac212
This synchronizes requirements.txt, test-requirements.txt and
setup.py with the current state of the openstack/requirements
global-requirements.txt list. Also adds flake8 exclusions to tox.ini
for new hacking rules with which git-review does not yet comply.
Change-Id: I837ba57134bea16dd46c3f512bccd51cb39f76c8
Simply skip the whole function when the remote has no reviews and print
out a friendly user message.
Change-Id: I4c07f05d4b797af056ee08016b150088f3b182db
Keep username/password defining and setting within the same module to
ensure any changes are kept in sync. Additionally remove the '--add'
option as git-review only supports one value for gitreview.username so
it should not be treated as a multi value config option.
Change-Id: I9ae64d8d1884e13a442da6f4d28407ca103e7b00
If the remote branch being targetted does not exist, any attempted
rebase will fail with an error that is largely unhelpful for users.
Assuming the preceding fetch succeeded properly, the most common reason
for the remote reference to not exist, is that the user wants the
changes to go to a new branch rather than the default. In this case
provide the user with a more helpful message indicating that if this is
the intention, to run git-review with the option to disable the rebase
in order to get the desired outcome of gerrit creating the branch
desired for the new reviews.
Change-Id: I4c92e901950abf2c879d00660daae3ccce7ddfc8
If commit message ends with "blueprint" or "bug" and Change-Id line
follows, git-review would set topic to "Change-Id". So I fixed the bug
Change-Id: Id1aadef6de622b57e475d698ab9a9f98c3428651
Tests will fail if users have a number of SSH identities in use
locally. Should prevent all but the created key from being offered
to the remote.
Change-Id: I13c915418af0740592a6328e3af330c38cdb0729
