A few days ago, I encountered frustrating behavior with hashset where hash is not defined for arbitrary ref objects, which as pointed out by Araq a few years ago (pre 2.0) is "a strange omission". The functionality has since been added behind a compiler flag, and from what I can tell it seems like it's seen as the more sensible default.
Given this, I was surprised to learn the change and flag were added 3 years ago and weren't moved to become default with 2.0.
Do we maintain a list of breaking changes we want to make? Is there a reason this wasn't included?
is the latest compiler/stdlib test suite runnable with old compiler version?
That might partially answer this question.
This particular change waits for https://github.com/nim-lang/RFCs/issues/380 which allows us to detect the resulting bugs.
The RFCs repository in general contains development plans.
Okay gotcha, so the RFC repo answers the title question (there's an candidate for v2 tag)
In the lifecycle of an RFC does "accepted" mean implementing this has been added to a backlog somewhere or that it's a topic that is open for community input?