Romain Marcadier

> Adding a section to the documentation on unit testing with examples. It would be super slick to have this integrated with the cli, ie. cdk run-tests This is interesting...

The interfaces should be declared as `Protocol`s, which means that normally, if they are correctly implemented, structural typing should make them match... Although I'd say I have seen cases where...

@Jacco one property of jsii structs is that they are pure-data, immutable objects. Adding setters to those would break the *immutable* property. So basically, we'd need those setters to be...

@rirze > Another quirk of the un-Pythonic `metaclass` requirement: invalid attributes (of a class that uses `JSIIMeta` as a metaclass) don't throw `AttributeErrors`. Instead they return quietly as None... Do...

@dmartin-gh > I am also experiencing the interface issues described above. We're looking to adopt the CDK for our main platform and as a mostly Python/C++ shop the Python CDK...

I'd be curious to know what service(s) was/were part of that $10K charge... I have some suspicions here, but I'd like to confirm them...

I wonder if we should reverse control here, and instead of having the kernel spawn the child process on behalf of the caller, it merely returned the `command`, `arguments`, and...

Those types cannot be represented in non-TypeScript languages. We could imagine allowing named types to reduce the boilerplate involved in re-implementing those types (`export type SomeName = Pick`), which `jsii`...

This would be addressed when we implement https://github.com/aws/aws-cdk-rfcs/pull/373, which we are currently planning.

Implementing inline diagnostic moderation would likely require a significant re-architecting of the compiler infrastructure... I think it is the right thing to do in the long run, but it will...