Creating a release

Releasing to PyPI

1. Gather closed issues and pull requests related to the release into a GitHub milestone:

   https://github.com/NeuralEnsemble/ephyviewer/milestones

2. Create a branch called "release-X.X.X" using the real release number.

3. In the docs, add a change log entry referencing each notable PR and issue (remember to add a link from releasenotes.rst!)

4. In version.py, set the new version number, making sure that you remove the ".dev".

5. Commit these changes to the branch with the message "Prepare release X.X.X", push the branch to GitHub, and open a pull request.

6. Do not merge yet, since you should first verify PyPI is happy with your changes. On your local machine, delete any old build directories:

   cd ephyviewer
   rm -rf build dist ephyviewer.egg-info
   

7. Build for distribution:

   python setup.py sdist bdist_wheel
   

8. Upload the build artifacts to PyPI's test server:

   twine upload --repository-url https://test.pypi.org/legacy/ dist/*
   

9. Verify the test release looks OK:

   https://test.pypi.org/project/ephyviewer/

10. If PyPI is OK, merge the pull request with commit name "Release X.X.X".

11. Verify that the new release notes built correctly:

    https://ephyviewer.readthedocs.io/en/latest/releasenotes.html

12. If the docs are OK, tag the PR merge commit with the release number and push the tag to GitHub.

13. If all is well, upload to the official PyPI server (THIS CANNOT BE UNDONE OR OVERWRITTEN):

    twine upload dist/*
    

14. Verify the official release looks OK:

    https://pypi.org/project/ephyviewer/

15. In version.py, increment the version number to the next patch release and append ".dev".

16. Commit these changes directly to master with the message "After release" and push to GitHub.

17. Create a GitHub release from the pushed tag containing the release notes:

    https://github.com/NeuralEnsemble/ephyviewer/tags

18. Close the milestone and create a new one for the next minor release:

    https://github.com/NeuralEnsemble/ephyviewer/milestones

Rebuilding for conda-forge

1. In your fork of conda-forge/ephyviewer-feedstock, create a branch called "build-X.X.X" using the real release number.

2. Edit the feedstock's meta.yaml file by:

   1. updating the version number
   2. updating the sha256 (which can be obtained for the tar.gz file here)
   3. resetting the build number to 0

3. If any other changes to the build recipe are required, make those now.

4. Commit these changes to the branch with the message "Build for release X.X.X", push the branch to GitHub, and open a pull request.

5. Post @conda-forge-admin, please rerender in the pull request and wait for the bot to update the recipe.

6. Make sure the checklist in the pull request's initial post is complete.

7. Wait for all tests to pass:

   https://dev.azure.com/conda-forge/feedstock-builds/_build?definitionId=8417&_a=summary

8. If tests are OK, merge the pull request.

9. The new build should appear on Anaconda Cloud fairly quickly:

   https://anaconda.org/conda-forge/ephyviewer/files

   but the index of the CDN that distributes the build may take a little longer to sync with it. The last time it was updated can be checked using this command:

   curl https://conda-static.anaconda.org/conda-forge/rss.xml -s 2>&1 | head -10 | tail -1

   If this takes a very long time (many hours), you can post on Gitter to see if anyone is available to force the CDN's index to update:

   https://gitter.im/conda-forge/conda-forge.github.io

Constructing installers

1. Update the NeuralEnsemble/ephyviewer-constructor recipe with the new version number, and any other required changes.

2. Build an installer for the current platform:

   cd ephyviewer-constructor
   constructor
   

3. Add the date to the installer (e.g., ephyviewer-1.2.0-2019.12.06-Windows-x86_64.exe).

4. Upload the installer to the new GitHub release page:

   https://github.com/NeuralEnsemble/ephyviewer/releases