What's the potential gain? This project has a laser sharp scope, it does one thing and it does that well. It's not like future development efforts are hampered by having some conditional macros in the C code. I agree that not many people should expect any support for Python 2 in 2021, so the downside is not that bad. I just can't see the what the project could gain from removing the option. Change my mind :)

Well, not sure if I'll find a good enough argument to convince you :)

It will become more difficult to test this library on Python 2 every year. Probably, in 5 years or so there will be no Python 2 pre-installed on MacOS and Linux. Breaking changes in some libraries will prevent you from building it from source (like OpenSSL 1.0 vs 1.1, it already makes it difficult to compile old Pythons). I agree that supporting Python 2 doesn't require any effort at the moment. It's not likely that something will change in Python 2. But Python 3 will definitely change and require updates. At some point, after a fix required for Python 3, you won't be even able to validate that it still works on Python 2. So, you'll have to drop it anyway sooner rather than later.