diff options
Diffstat (limited to 'third_party/python/coverage/howto.txt')
| -rw-r--r-- | third_party/python/coverage/howto.txt | 122 |
1 files changed, 122 insertions, 0 deletions
diff --git a/third_party/python/coverage/howto.txt b/third_party/python/coverage/howto.txt new file mode 100644 index 0000000000..3653e830a7 --- /dev/null +++ b/third_party/python/coverage/howto.txt @@ -0,0 +1,122 @@ +* Release checklist + +- Check that the current virtualenv matches the current coverage branch. +- Version number in coverage/version.py + version_info = (4, 0, 2, 'alpha', 1) + version_info = (4, 0, 2, 'beta', 1) + version_info = (4, 0, 2, 'candidate', 1) + version_info = (4, 0, 2, 'final', 0) +- Python version number in classifiers in setup.py +- Copyright date in NOTICE.txt +- Update CHANGES.rst, including release date. + - don't forget the jump target +- Update README.rst + - "New in x.y:" + - Python versions supported +- Update docs + - Python versions in doc/index.rst + - Version of latest stable release in doc/index.rst + - Version, release, release_date and copyright date in doc/conf.py + - Look for CHANGEME comments + - Don't forget the man page: doc/python-coverage.1.txt + - Check that the docs build correctly: + $ tox -e doc + there will be warnings about the readthedocs links being broken, + because this version's docs haven't been published yet. + - Done with changes to source files, check them in. + - git push + - Generate new sample_html to get the latest, incl footer version number: + make clean + pip install -e . + cd ~/cog/trunk + rm -rf htmlcov + coverage run --branch --source=cogapp -m pytest -k CogTestsInMemory; coverage combine; coverage html + - IF PRE-RELEASE: + rm -f ~/coverage/trunk/doc/sample_html_beta/*.* + cp -r htmlcov/ ~/coverage/trunk/doc/sample_html_beta/ + - IF NOT PRE-RELEASE: + rm -f ~/coverage/trunk/doc/sample_html/*.* + cp -r htmlcov/ ~/coverage/trunk/doc/sample_html/ + cd ~/coverage/trunk + - IF NOT PRE-RELEASE: + check in the new sample html + - Build and publish docs: + - IF PRE-RELEASE: + $ make publishbeta + - ELSE: + $ make publish +- Kits: + - Start fresh: + - $ make sterile + - Source kit and wheels: + - $ make kit wheel + - Linux wheels: + - $ make kit_linux + - Windows kits + - wait for over an hour for Appveyor to build kits. + - https://ci.appveyor.com/project/nedbat/coveragepy + - $ make download_appveyor + - examine the dist directory, and remove anything that looks malformed. + - test the pypi upload: + - $ make test_upload +- Update PyPI: + - upload kits: + - $ make kit_upload +- Tag the tree + - git tag coverage-3.0.1 + - git push --tags +- Bump version: + - coverage/version.py + - increment version number + - IF NOT PRE-RELEASE: + - set to alpha-0 if just released. + - CHANGES.rst + - add an "Unreleased" section to the top. + - git push +- Update Tidelift: + - make upload_relnotes +- Update readthedocs + - IF NOT PRE-RELEASE: + - update git "stable" branch to point to latest release + - git branch -f stable <latest-tag> + - git push --all + - visit https://readthedocs.org/projects/coverage/builds/ + - wait for the new tag build to finish successfully. + - visit https://readthedocs.org/dashboard/coverage/advanced/ + - change the default version to the new version + - visit https://readthedocs.org/projects/coverage/versions/ + - find the latest tag in the inactive list, edit it, make it active. + - readthedocs won't find the tag until a commit is made on master. + - keep just the latest version of each x.y release, make the rest inactive. +- Visit the fixed issues on GitHub and mention the version it was fixed in. + - make a milestone for the next release and move open issues into it. +- Announce: + - twitter @coveragepy + - nedbatchelder.com blog post? + - testing-in-python mailing list? + + +* Testing + +- Testing of Python code is handled by tox. + - Create and activate a virtualenv + - pip install -r requirements/dev.pip + - $ tox + +- Testing on Linux: + - $ make test_linux + +- For complete coverage testing: + + $ make metacov + + This will run coverage.py under its own measurement. You can do this in + different environments (Linux vs. Windows, for example), then copy the data + files (.metacov.*) to one machine for combination and reporting. To + combine and report: + + $ make metahtml + +- To run the Javascript tests: + + open tests/js/index.html in variety of browsers. |