Is there a reason why we are dropping the type hint here?

I don't remember exactly. I thin I got a cyclic import error at some point due to this import. I can check again to see if we really need to remove it.

So, I just checked and we need to remove it to avoid a cyclic import error

mypy/typing actually has a solution for just this problem! https://mypy.readthedocs.io/en/stable/common_issues.html#import-cycles

tl;dr:

from typing import TYPE_CHECKING

if TYPE_CHECKING:
    from poetry.puzzle.provider import Provider

Fixed!

I can see crashtest on pypi but not on github, is this expected?

It's not open source yet, I need to write a proper README for the project first. It should be ready tomorrow.

It's now open source: https://github.com/sdispater/crashtest

These modules are a mouth full :) Would be great to put them into a structure, eg poetry.solutions.providers.incompatibility.python_requreiment.

Would solutions be better placed in poetry.mixology consdiering we only really provide solutions for version solving issues or is the intent here to provide a generic mechanism?

I wanted them to be as explicit as possible :-) But yeah I can see if we can reorganize a bit.

The idea is to extend the solutions beyond solver issues, for instance for configuration and validation errors.

So, I reorganized a bit and now each module (in this case mixology) can provide its own solutions and providers.

This allows for a more flexible implementation.