This documentation covers IPython versions 6.0 and higher. Beginning with version 6.0, IPython stopped supporting compatibility with Python versions lower than 3.3 including all versions of Python 2.7.
If you are looking for an IPython version compatible with Python 2.7, please use the IPython 5.x LTS release and refer to its documentation (LTS is the long term support release).
Guide for IPython core Developers¶
This guide documents the development of IPython itself. Alternatively, developers of third party tools and libraries that use IPython should see the Developer’s guide for third party tools and libraries.
For instructions on how to make a developer install see Installing the development version.
Backporting Pull requests¶
All pull requests should usually be made against
main, if a Pull Request
need to be backported to an earlier release; then it should be tagged with the
If you tag a pull request with a milestone before merging the pull request,
and the base ref is
main, then our backport bot should automatically create
a corresponding pull-request that backport on the correct branch.
If you have write access to the IPython repository you can also just mention the backport bot to do the work for you. The bot is evolving so instructions may be different. At the time of this writing you can use:
@meeseeksdev[bot] backport [to] <branchname>
The bot will attempt to backport the current pull-request and issue a PR if possible.
[bot] when mentioning the bot should be optional and can
If the pull request cannot be automatically backported, the bot should tell you so on the PR and apply a “Need manual backport” tag to the origin PR.
IPython release process¶
This document contains the process that is used to create an IPython release.
release script in the
tools directory of the
repository automates most of the release process. This document serves as a
handy reminder and checklist for the release manager.
During the release process, you might need the extra following dependencies:
keyringto access your GitHub authentication tokens
graphvizto generate some graphs in the documentation
ghproto generate the stats
Make sure you have all the required dependencies to run the tests as well.
You can try to
source tools/release_helper.sh when releasing via bash, it
should guide you through most of the process.
1. Set Environment variables¶
Set environment variables to document previous release tag, current release milestone, current release version, and git tag.
These variables may be used later to copy/paste as answers to the script
questions instead of typing the appropriate command when the time comes. These
variables are not used by the scripts directly; therefore, there is no need to
export them. The format for bash is as follows, but note that these values
are just an example valid only for the 5.0 release; you’ll need to update them
for the release you are actually making:
PREV_RELEASE=4.2.1 MILESTONE=5.0 VERSION=5.0.0 BRANCH=main
For reproducibility of builds,
we recommend setting
SOURCE_DATE_EPOCH prior to running the build; record the used value
SOURCE_DATE_EPOCH as it may not be available from build artifact. You
should be able to use
date +%s to get a formatted timestamp:
2. Create GitHub stats and finish release note¶
This step is optional if making a Beta or RC release.
Before generating the GitHub stats, verify that all closed issues and pull requests have appropriate milestones. This search should return no results before creating the GitHub stats.
If a major release:
merge any pull request notes into what’s new:python tools/update_whatsnew.py
docs/source/whatsnew/development.rst, to ensure it covers the major release features
move the contents of
Xis the numerical release version
generate summary of GitHub contributions, which can be done with:python tools/github_stats.py --milestone $MILESTONE > stats.rst
which may need some manual cleanup of
stats.rst. Add the cleaned
Xis the numerical release version (don’t forget to add it to the git repository as well). If creating a major release, make a new
github-stats-X.rstfile; if creating a minor release, the content from
stats.rstmay simply be added to the top of an existing
docs/source/whatsnew/index.rstto list the new
github-stats-Xfile you just created.
You do not need to temporarily remove the first entry called
development, nor re-add it after the release, it will automatically be hidden when releasing a stable version of IPython (if
release.pyis an empty string.
Make sure that the stats file has a header or it won’t be rendered in the final documentation.
To find duplicates and update
git log --format="%aN <%aE>" $PREV_RELEASE... | sort -u -f
If a minor release you might need to do some of the above points manually, and forward port the changes.
3. Make sure the repository is clean¶
- of any file that could be problematic.
Remove all non-tracked files with:
git clean -xfdi
This will ask for confirmation before removing all untracked files. Make sure the
dist/folder is clean to avoid any stale builds from previous build attempts.
4. Update the release version number¶
IPython/core/release.py to have the current version.
in particular, update version number and
_version_extra content in
Step 5 will validate your changes automatically, but you might still want to make sure the version number matches pep440.
beta are not separated by
. or the
bdist will appear as different releases. For example, a valid version
number for a release candidate (rc) release is:
1.3rc1. Notice that there
is no separator between the ‘3’ and the ‘r’. Check the environment variable
$VERSION as well.
You will likely just have to modify/comment/uncomment one of the lines setting
5. Run the
tools/build_release does all the file checking and building that
the real release script will do. This makes test installations, checks that
the build procedure runs OK, and tests other steps in the release process.
build_release script will in particular verify that the version number
match PEP 440, in order to avoid surprise at the time of build upload.
We encourage creating a test build of the docs as well.
6. Create and push the new tag¶
Commit the changes to release.py:
git commit -am "release $VERSION" -S git push origin $BRANCH
-S if you are no signing the package)
Create and push the tag:
git tag -am "release $VERSION" "$VERSION" -s git push origin $VERSION
-s if you are no signing the package)
Update release.py back to
x.y-maint commit and push:
git commit -am "back to development" -S git push origin $BRANCH
-S if you are no signing the package)
Now checkout the tag we just made:
git checkout $VERSION
7. Run the release script¶
release script, this step requires having a current wheel, Python
>=3.4 and Python 2.7.:
This makes the tarballs and wheels, and puts them under the
folder. Be sure to test the
wheels and the
sdist locally before
uploading them to PyPI. We do not use an universal wheel as each wheel
ipython3 script, depending on the version of
Python it is built for. Using an universal wheel would prevent this.
Check the shasum of files with:
shasum -a 256 dist/*
and takes notes of them you might need them to update the conda-forge recipes. Rerun the command and check the hash have not changed:
./tools/release shasum -a 256 dist/*
Use the following to actually upload the result of the build:
It should posts them to
archive.ipython.org and to PyPI.
PyPI/Warehouse will automatically hide previous releases. If you are uploading a non-stable version, make sure to log-in to PyPI and un-hide previous version.
8. Draft a short release announcement¶
The announcement should include:
a link to the html version of the What’s new section of the documentation
a link to upgrade or installation tips (if necessary)
Post the announcement to the mailing list and or blog, and link from Twitter.
If you are doing a RC or Beta, you can likely skip the next steps.
9. Update milestones on GitHub¶
These steps will bring milestones up to date:
close the just released milestone
open a new milestone for the next release (x, y+1), if the milestone doesn’t exist already
10. Update the IPython website¶
The IPython website should document the new release:
add release announcement (news, announcements)
update current version and download links
update links on the documentation page (especially if a major release)
11. Update readthedocs¶
Make sure to update readthedocs and set the latest tag as stable, as well as checking that previous release is still building under its own tag.
12. Update the Conda-Forge feedstock¶
Follow the instructions on the repository
Celebrate the release and please thank the contributors for their work. Great job!
Out of date documentation is still available and have been kept for archival purposes.
Developers documentation used to be on the IPython wiki, but are now out of date. The wiki is though still available for historical reasons: Old IPython GitHub Wiki.