Requires-Python from Wheel needs audit
I believe we need to clean it before we build the wheel
Forgot to mention but I experiemented a bit, and here’s the important part
https://github.com/sarugaku/passa/blob/eb7194dd2ce3650d5dd902b8671515d034be5570/src/passa/dependencies.py#L192-L210
This is almost definitely not covering all cases, but I couldn’t find many good 1.x wheel examples, so we’ll have to fix it as we go.
Also I found that even if I’m reading the correct field, JSON API and the wheel can still return different values. This is just terrible.
oh yeah to summarize what we talked about whenever that was, and the reason why we have the custom cleaning stuff in pipenv, Requires-Python: 3 is completely valid for example
Indeed. This function (or the resolver in general) does not try to actually parse the string in any sense. The parsing part is done after the resolution, by the marker thing you implemented.