- Promote Paul Ganssle
- Do not promote Paul Ganssle
Victor Stinner and I want to propose Paul Ganssle as a core developer, who would center his efforts on co-maintaining datetime together with Alexander Beloposky.
Some of you may already know Paul Ganssle as the main maintainer of dateutil, as maintainer of setuptools or for his contributions to CPython. On the technical side, Paul has proven many times to have a great domain knowledge around the datetime module and related functionality but he also has remarkable knowledge about Python and CPython internals. As a reviewer, Paul has made several reviews of datetime-related pull requests but also other general pull requests involving Python
and C code as well as documentation. In the reviews he always shows a great care towards the contributor (as can be seen as well in the reviews in the packages that Paul maintains) but also he has spend a lot of time reviewing the wording as a native English speaker, always in a detailed and pedagogical way. Paul always listens to other ideas and viewpoints when discussing and he reacts very positively to criticism in his PRs and work.
Paul would like to focus on co-maintain datetime together with Alexander Beloposky (who has expressed that he is ok with that).
We consulted with Paul prior to the nomination, and he said he is interested in becoming a core developer so that he can expand his ability to improve the datetime module and CPython in general; he also participates in many sprint events, and it would improve his ability to bring new contributors on to the project. In short, we think Paul is an exceptional developer, a kind and compassionate person, and CPython would benefit from having him on the core team.
As a reminder from PEP 13 regarding voting rules:
… It is granted by receiving at least two-thirds positive votes in a core team vote that is open for one week and with no veto by the steering council."
Here is a summary of Paul’s work and achievements:
Background Information about Paul Ganssle
- Maintainer of dateutil since 2014. First and most prominent third-party implementation of PEP 495 (Local Time Disambiguation).
- Maintainer of setuptools since 2018.
- Frequently run sprints on dateutil and setuptools. Would like to be able to offer mentorship on CPython at conference sprints.
- Organized and ran the PyPA mini-summit at PyCon US 2018 and 2019, active in the PyPA generally.
- Wrote datetime bindings for PyO3 (CPython bindings for Rust), and has continued to contribute regularly to that project with PRs, reviews and issues. Interested in contributing to the C API from the perspective of someone writing cross-language bindings.
- Frequently helps maintain datetime-related code on other projects:
- Spoke at the language summit about adding more time zones to the standard library. Slides, Repo
Major accomplishments in CPython
datetime.fromisoformat, the inverse of
datetime.isoformat. This is still the fastest constructor for
datetimeaccessible from Python. (PR #4699, PR #8959)
datetime.timezone.utcin the CPython API. (PR #5032)
- Implemented and made the case for changing the return type of
timedeltaarithmetic to respect the subclass of the
datetimeobject. Similarly has been working to make
datetimesafer to subclass in general. (PR #10902, Python-dev thread 1 2)
- Improved the speed of all
datetimealternate constructors. (PR #4993)
datetime.fromisocalendar, the inverse function from
datetime.isocalendarin PR #11888. Intends to continue improving and maintaining symmetry between “serialization” and “deserialization” functions.
General maintenance tasks
- Generally follows and comments on datetime and packaging related issues in the issue tracker, among other things.
- Reviews PRs, generally to datetime and time, but also unrelated PRs (e.g. PR #7605 adding
Also has been trying to help give a native English speaker’s perspective on documentation PRs.
During our discussion, Paul has provided a list of improvements to Python he’d like to make:
- Improve Python datetime C API test suite and documentation (partially completed at the PyCon sprints, Paul created the issues and helped Edison Abahurire with reviews and in-person guidance while he made his first pull requests adding tests and documentation for the datetime C API).
- Expand time zone support in the standard library with a concrete implementation of IANA time zones and an explicit representation of “local” time.
- Add nanosecond support to
- Improve compatibility across platforms where possible in things like
- Improve locale-specific testing
Most of this is datetime related because that is the primary area where his expertise is most needed (plus most packaging changes occur on
setuptools itself and not
distutils), but he is interested in general contributions to the language as well.
- PRs authored: https://github.com/python/cpython/pulls?utf8=✓&q=is%3Apr+author%3Apganssle
- PRs reviewed: https://github.com/python/cpython/pulls?utf8=✓&q=is%3Apr+reviewed-by%3Apganssle+-author%3Apganssle
Selected bpo issues
strftime() returns wrong week numbers: https://bugs.python.org/issue35841
warning about potential dead code: https://bugs.python.org/issue36330
Wrong inspect.getsource for datetime: https://bugs.python.org/issue32313
fromisoformat doesn’t handle “Z”: https://bugs.python.org/issue35829
sqlite3 optional auto-conversion of DATETIME fields: https://bugs.python.org/issue35145
Incorrect documentation for strftime(): https://bugs.python.org/issue33381
Replace append loops with comprehensions: https://bugs.python.org/issue36039
datetime.time.isoformat has inconsistent behavior: https://bugs.python.org/issue34407