Currently, the repository contains 4 Cython-generated C source files. I think there are a few good reasons to remove these files:
This thread gives a variety of opinions on whether generated files should be committed or not. The consensus is that there are some niche reasons to do this (e.g. the hand-written code is tightly coupled to the generated code, or the code generator is unreliable), but generally you shouldn't. I don't think any of the niche reasons apply here.
Generated files can get out-of-date, and interfere with installation in confusing ways (as described in #26).
Any commit made with a different version of Cython will have really big diffs. This obfuscates what actually changes in each commit, and it can easily lead to merge conflicts as well.
I can think of one possible reason to not remove these files, which is that they make the software more reproducible by ensuring that everyone gets the same generated source. But I don't think this is a very compelling reason:
Without frequent updates (e.g. 2-3x per year), the generated source files get out-of-date and lead to the installation issues mentioned above. Frankly, this repository is just not updated frequently enough for this to work.
Distributing the generated source files doesn't require checking them into the repository. It just requires including those files in the archive uploaded to PyPI.
I expect that merging either this PR or #30 will cause merge conflicts for the other. If/when this happens, I'll fix the conflicts.
Currently, the repository contains 4 Cython-generated C source files. I think there are a few good reasons to remove these files:
I can think of one possible reason to not remove these files, which is that they make the software more reproducible by ensuring that everyone gets the same generated source. But I don't think this is a very compelling reason:
I expect that merging either this PR or #30 will cause merge conflicts for the other. If/when this happens, I'll fix the conflicts.