Jon Pretty

Yep. It's a complete&utter trainwreck. Having made multiple enum impls over the years, I'd have died in shame proposing something like that.

Is the problem that they don't quite offer a complete replacement for sealed traits and case classes?

I think I found 6 or 7 design mistakes. Each significant enough to outright reject this proposal. But SIP process doesn't apply anyway.

Not to Dotty, it doesn't... but Martin did change the proposal in response to feedback on his PR, so it's not like it was a closed process.

I don't think any changes can fix it. It's not only what it does, but that the whole thing is a confused, sprawling, over-complicated mess.

I think the risk is that it provides concise support for something you can already do, and costs one feature, and less syntax uniformity.

Thanks - that's a more constructive teardown, even though I don't agree with all of it. :)