Yeah, the names imply that those should be TZ aware. Though I can understand the reasoning, as this change can silently break already broken code that kind of works, and would be a nightmare to fix in a large codebase.

I also disagree with author, about naive datetime. I don't think there is a good use case for them, and people who use them, do so because they are not aware of the issues.

The only argument I saw was for micro performance improvements, which is kind of funny to care about when programming in python. Since it is implemented in C, the code removing the timezone likely takes more resources.