* Update couchbase install steps
Fix integration test for channel leak. We have to inspect the channels created before the tasks are done to ensure we're looking at the correct data
Running multiple times test_parallel_chords to make sure test works
Add pytest fixtures to t/integration/conftest.py so we don't have to install package to use fixtures
Update sphinx test to use sphinx_testing instead of running sphinx-build
Bump setuptools
Install reqs for all python versions and upgrade pip
Update docker-compose to create a tagged image and update PYTHONPATH
Add bandit to pkgutils
Update contributing documentation and changelog to show 4.3 version
Add pkgutils and docs requirements to run sphinx unit tests and use cyanide when running travis
Forgot to fix flake8 issues on tests. Remove bandit from pkgutils.txt since tox already installs it. Update CONTRIBUTING.rst to show how to install bandit to run it
Fix flake8 issues on test_sphinx and add shared task to the test
Update wording for CONTRIBUTING.rst
Make python3.6 default python version, mount the entire celery folder so everything can be done inside the container and bump setuptools
Update label definitions.
Remove cyanide from requirements for now and add bumpversion information.
* Update celery.contrib.sphinx. Checking if the object to document is a subclass of
BaseTask and has the attribute __wrapped__ should be enough to know if it's a Celery task.
Checking if the object is also an instance of Proxy/PromiseProxy makes the extension not work correctly.
Probably becuase of how sphinx loads objects and the dunder overrides that the Proxy class does,depending on how sphinx-doc is ran a celery task might be or not a
instance of Proxy.
* Update Test Case details.
* Added docs requirements to Docker image
* sphinx_celery requires Sphinx 1.7.1
pip-9.0.1 didn't complain about an older version being installed, but
pip-10.0.1 does.
This is achieved by cleaning up pyc files in the docker entrypoint
script which avoids mismatches between compiled Python files on the
docker host and/or from previous build steps by construction: before
executing any command in the docker container, any remnant platform
specific files are cleaned up and as such any subsequent py.test command
will run in a clean environment.
Running docker-compose on a clean checkout currently fails with a
validation error that `services.celery.build.args` contains null.
Referencing the `CELERY_USER` environment variable defined in `.env`
does not fix this since environment variables are only applied at
container run-time, not build-time. To avoid potential confusion due to
this, it's easiest to just hard-code the `CELERY_USER` variable in the
docker-compose file.
* Add docker-compose and base dockerfile for development
* Change to base jessie docker image and setup pyenv
* Add in aliases for pyenv, fix entrypoint for docker
* Update dockerfile to be non-root and fix encoding problems with tox
* Add convenience method to get redis connection for tests
* Add pypy to install commands
* Force worker to pick up broker url from environment
* Move pypy comment to above apt-get install, default python to 3.6, add in flag to prevent bytecode
* Add documentation
* Fix links
* Update docs
* Add to contributors
* Address feedback: improve documentation, separate dockerfile into scripts, remove redundancy in pyenv setup, add in .env file
* Setup pyenv environments correctly in dockerfile
* Update capitalization
* Change CELERY_USER to ARG in dockerfile and pass build argument in build
* Change default worker loglevel to debug in docker-compose
Omer Katz