deprecate "installer.modern-installation false"

[dimbleby]

As proposed at #8987

I started with warnings.warn but something about all the cleo io cleverness results in those messages disappearing. Anyway this deprecation is more user-facing than developer-facing, so regular warning output is probably fine.

I'd be happy to see this deprecation in the upcoming release, might as well get it out there and see who squeals. But no big deal if it misses the cut.

[radoering]

One thing: Shouldn't we use write_error_line() for warnings?

[dimbleby]

some do, some don't: if there is a rule then it is not consistently followed. The stdout from these commands is not really suitable for parsing anyway so I don't suppose it much matters.

I'm not at the right computer to do anything about this now, feel free do any of tweak it yourself / ask me to do it later / not bother.

[radoering]

You are probably right that it is inconsistent and does not matter.

Interestingly enough, I see random test failures on Windows if I write the warning to error out. It's always a PermissionError in

poetry/tests/conftest.py

Line 356 in 0eedea4

shutil.rmtree(venv.path)

but in different tests, for example one or two of

tests/console/commands/self/test_add_plugins.py::test_add_with_constraint
tests/console/commands/self/test_add_plugins.py::test_add_with_git_constraint_with_extras
tests/console/commands/self/test_remove_plugins.py::test_remove_installed_package
tests/console/commands/self/test_remove_plugins.py::test_remove_installed_package_dry_run
tests/installation/test_executor.py::test_executor_should_write_pep610_url_references_for_editable_git

Maybe, a reason to just keep it as is, even if this is a bit unsettling...

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.