@hroncok My counterpoint here:

- The current Python REPL is pretty rough, and this is a *dramatic* improvement.
- A lot of the weird behavior and bugs don't show up until you get a lot of eyes on it, which means that the first release will always look like this.
- Given that it's an interactive terminal and not part of "hands-off" deployments, the kinds of bugs you are likely to encounter are less like, "Oh this created a bunch of work for me today because our whole pipeline broke" and more like, "Oh this new REPL isn't working for me, I should disable it until it works".

I am neutral about the lack of a PEP. It's definitely a major new feature, which weighs in favor of a PEP, but also I'm not sure how well it fits with the PEP process. It's not a major language feature, it's more an overhaul of an interactive application bundled with the language. I'm not sure what choices or trade-offs needed to be made that they would want people to weigh in about, or what decisions the SC would be making.

I also think the case could be made for saying that if you were to make the individual changes one at a time, none of them would require a PEP (with the possible exception of `exit() -> exit`).