Adopt pyproject.toml

[ThomasWaldmann]

Modern python packaging expects to find a pyproject.toml, which can also be used to contain configuration information for a lot of python-related tools.

Direct calls to setup.py are deprecated. Fixes #1619.

Fixes #1571.

[ThomasWaldmann]

@RogerHaase I put your name/email into maintainer list (as found in git logs).

@UlrichB22 I didn't find a usable email in git logs, so I didn't add you into maintainer list yet.

Are you both ok with adopting ruff instead of flake8? The latter still can't read its settings from pyproject.toml.

[ThomasWaldmann]

Note: I didn't do much practical testing of this yet, help is appreciated.

[UlrichB22]

I started testing in a fresh install from a git clone (for developers).

Following tests were successful after short check:

• quickinstall.py
• ./m extras
• ./m tests
• moin create-instance --full && moin run

Now I'm testing pybabel extract, the resulting MoinMoin.pot is only half the size of the old version and git diff doesn't help because of the different sort order. But that's not a showstopper. I will investigate this further as I want to update the German translations in the next step.

For me it is fine not to be on the list of maintainers.

[ThomasWaldmann]

@UlrichB22 thanks for testing.

translations: maybe the size difference comes from (missing) comments? iirc, there is an option to include src file / line no as a comment.

[UlrichB22]

Regarding babel, please change babel.cfg to

[python: **/**.py]
encoding = utf-8

[jinja2: **/templates/**.html]
encoding = utf-8

[jinja2: **/templates/dictionary.js]
# all JS translatable strings must be defined here for jQuery i18n plugin
encoding = utf-8

and add following options to the babel extract command in docs/devel/translate.rst:

--project="moin" 
--version="<version>"

[ThomasWaldmann]

When reviewers think this is ready, please approve.

After that, I'ld like to squash some commits before merging.

[RogerHaase]

I tested clone on windows, all OK except for test 3 errors (Cannot create a file when that file already exists:) that did not repeat on 2nd try. This happens occasionally on Windows, workaround is to rerun.

[ThomasWaldmann]

@RogerHaase Are you saying this problem is new, caused by my changes and does not happen in master branch?

[RogerHaase]

Not a new problem. Windows seems to have problems dealing with tests that delete a file and then recreate a file with same name. Happens around 1 in 50 or 100 tests. Cannot reproduce.

[ThomasWaldmann]

@RogerHaase OK, guess that should be in a ticket then, not in this PR. I remember we have seen very strange windows fs effects in the past, too.

[UlrichB22]

The translation with pybabel works fine now.
I think you will test all the packaging stuff.

[ThomasWaldmann]

I manually and automatically compared the sdist *.tgz files produced by this PR and master branch - there were no unexpected differences.

So, if there is nothing else, I'll squash some commits and make it ready to merge.

[ThomasWaldmann]

Squashed some fix commits and force-pushed. Ready AFAIAC.