Closed ajstewart closed 2 years ago
@marxide is this worth it?
If so I kind of feel like the user guidance to install using pip
should also be removed and replaced with poetry
, but I can't help but feel this is slightly less user friendly?
I tried to search about using pip
for poetry
managed projects and if it respects the lock file at all but I couldn't find much.
Maybe having the pipeline pip install
-able is incompatible with wanting to use lock files. I couldn't find much on the topic either, but this SO answer recommends building wheels with poetry and then installing those with pip: https://stackoverflow.com/a/67534734. Our tests could do this automatically. For users, we could upload wheels when we make releases, which would maintain the current pip install
compatibility. If users want to reliably build/install from source, they'll need Poetry.
Whatever we do, I agree that the tests should replicate what our docs tell users to do.
Yeah I'm not sure what the best method is here. Especially as it's not really a 'package' as such, it's purely only building dependencies, so setting it up for a pip install <package>
doesn't seem that intuitive either. Is requiring users to use poetry a bad thing?
Decided not to pursue.
Changes the use of
pip
in the test-suite to install the dependencies topoetry
.