build(medcat): Add support to latest build systems.#568
Conversation
There's been a release of setuptools_scm (10.2.0) that fixed the underlying issue. Unfortunately there was no mention of the issue in the [release notes](https://github.com/pypa/setuptools-scm/releases/tag/setuptools-scm-v10.2.0). However, the [fix PR](pypa/setuptools-scm#1441) commit (5b640c6) is within the [compare](pypa/setuptools-scm@setuptools-scm-v10.1.2...setuptools-scm-v10.2.0) so I am fairly certain the fix is in place. But just to make sure, I ran the build locally as well and verified the build was successful and that no parent folders were being moved (which was the case previously). So all in all, what this PR does is fix the offending packages ( and ) to not include the versions within which they broke the builds.
alhendrickson
left a comment
There was a problem hiding this comment.
I'm good to approve, it does feel like the right thing to explicitly exclude broken versions. Like it's 100% correct.
But could you instead just bump the minimum version ? Just so we dont have to keep these in the file forever. Like can we just do setuptools_scm>=10.2.0 ? Or does this break downstreams
Good point. I was too stuck in avoiding things that doesn't work! I don't know of any reason we can't just bump the minimums. Newer stuff would be used anyway. |
There's been a release of setuptools_scm (10.2.0) that fixed the underlying issue. Unfortunately there was no mention of the issue in the release notes. However, the fix PR commit (5b640c6) is within the compare so I am fairly certain the fix is in place.
But just to make sure, I ran the build locally as well and verified the build was successful and that no parent folders were being moved (which was the case previously).
So all in all, what this PR does is fix the offending packages ( and ) to not include the versions within which they broke the builds.