PEP 541 Request: html5lib

[hugovk]

Project to be claimed

PROJECT_NAME: https://pypi.org/project/html5lib

Your PyPI username

USER_NAME: https://pypi.org/user/hugovk

Reasons for the request

The project appears to have been abandoned.

• Last release: Dec 2017 https://pypi.org/project/html5lib/#history

The last commit was in October 2019 but there's been no reply about a new release since May 2018.

The library will break in Python 3.10 due to the collections.abc deprecation removals.

It is used by many projects, for example it's vendored by pip (itself vendored in CPython's ensurepip module), by Bleach, and blocks Poetry running on new Python with the removal.

Due to the lack of release, the collections.abc deprecation removal was delayed from CPython 3.8 to 3.9, and this week has since been delayed again to 3.10. I expect the problem to resurface when work begins on 3.10 in or around May 2020.

In May 2018, @pganssle requested a new release. In December 2019, @vstinner emailed the maintainers @jgraham or @gsnedders, and has had no reply:

• New release on PyPI? html5lib/html5lib-python#419

Maintenance or replacement?

Maintenance from a fork: https://github.com/html5lib-python/html5lib-python. I've commented on open PRs and approved some. I've merged those to my fork, and would plan a release to unblock the collections.abc issue, and appoint other community maintainers.

Contact and additional research

No replies to html5lib/html5lib-python#419 or to Victor's email.

[jamadden]

@gsnedders has begun responding to the issue in the project repo after this request was entered, so I think this can be closed.

[hugovk]

Yes, good news, we don't need this now. Thanks all!

[jayaddison]

From my (limited) perspective there's been a good resurgence in activity on html5lib during the summer thanks to the work of @gsnedders. That said I still have concerns that the library may be encountering some difficulties maintaining review, improvement and release momentum.

Perhaps that's fine since html5lib is vendored within pip and thus stability is of very high importance, but other projects do manage to retain velocity while retaining safety. It's possible html5lib could benefit from some kind of support community.

[mgorny]

@hugovk, do you happen to be still interested in this? html5lib is unmaintained again, and now html5lib-modern fork has sprung (which is basically a one-time bugfix rather than something intended for long-term maintenance, see RDFLib/rdflib#2935) which is causing problems for packaging now. It would really be preferable to have a maintained first-party html5lib, so people wouldn't be working around its lack of maintenance.