jhugman

> This seems to imply that this problem lives in the domain of the foreign binding generators? I think to a first approximation this analysis is correct; though I'm still...

For reference, we had a similar issue in the FML, for rendering object literals that needed access to the Object definition. In that solution, [we defined a `LiteralRenderer` which could...

I'm not sure there's an overall plan to keep the oracle stateless. The original intention of the oracle was to contain both the definitions and the code types; that oracles...

For android, this is implemented as `androidx.annotation.VisibleForTesting`, but the annotation processor is part of androidx, rather than kotlin itself. This will unlikely to be a problem for uniffi's consumers, but...

Thanks both. ### Naming I can definitely get behind `Decorator`, which I'd have got from a Patterns book. The full pattern give the decorator access to the call arguments so...

> Also, thanks for making the delegates.md file. It's great to see the docs written with the code and even better to have examples to discuss. Once again, I see...

Raised #1054 to divert async/threading discussions there. While this delegates/decorators may be useful for some async usages, it's not its primary motivation. Analogy: delegates is to async, as `for` loops...

@mhammond suggested rewriting a couple of components in terms of decorators. [Rewriting Nimbus][1] and FxAClient, I'm seeing reductions of Kotlin file size by around half. The best bit is that...

> I'm torn because this code as-is would make life better for the Nimbus devs. FWIW, as Mark suggested, I [re-factored `Fxaclient` in terms of a decorator][1]; I didn't go...

> The tl;dr is that IMO, udl is doing us amazing service, but ultimately is more liability than asset […] - so not entrenching it further seems like a good...