Romain Marcadier

That would indeed be very useful. Alternatively, providing a way to write-if-not-exists, that could possibly be extended to write-unless-newer-than would bridge the gap as well.

Hey @Tinche - sorry for taking so long to get back to you! The unstructured type looks like so: ```py @attr.s(auto_attribs=True, frozen=True, slots=True) class SetRequest: objref: ObjRef property: str value:...

Investigating further on this problem, it appears the `value` field here cas actually unstructured *BEFORE* it is set on the `SetRequest` value, and this actually gives me more ways to...

I found that installing `libstdc++6` (instead of `libstdc++`) fixed the warnings in my case.

@skinny85 thanks for your feedback! I'm providing some answers below... --- > In my opinion, forcing casting things to `any` and using parameter overrides is too big of a drop...

> > > I would bet if I did a similar search for TypeScript, I would get ten times as many results. So this capability is actively being used by...

> Potentially we can [start with] disallowing object literals from being sent out by JS where the type is declared as a union. @rix0rrr - this would be a breaking...

@onefuncman - You had objected on the initial premise... It's been changed and I think you might like it better :)

Yeah - we are in this situation largely because we're still on TypeScript 3.9, and most of the ecosystem around us is gradually adopting TypeScript 4.3 to 4.5 minimum targets...

> Would setting skipLibCheck to true prevent jsii from looking in node_modules to begin with? I'm not sure... it could have other consequences on the TypeChecker AST we obtain during...