Ryan Kelly

Flagging this as a potential good-first-bug, if we can provide a few more details.

Nice work on debugging this! It's not 100% clear to me what change we should be making in a-s to help with this, should we e.g. be throwing an error...

Semi-answering my own question...so IIUC, `subscribe` currently lazily-initializes a `uaid` for the push manager when a subscription is created for the first time, but we should either do this earlier...

I'm flagging this for "discuss" because it seems like we're very close to understanding what's going on here and making it actionable.

Thanks for the suggestion! > If subclassing is not required I don't think we made an explicit decision on this front, but it turns out we *accidentally* decided this for...

> Best I can tell, uniffi will have the exact same problem described (and fixed) there. IIRC, my expectation at the time was that UniFFI would not have this problem...

(Previously we would define an exception named `Error`, which also shadows a Java buitlin of the same name, but we don't use that name in our generated code anywhere so...

Also JSON-related datatypes, ref https://github.com/mozilla/uniffi-rs/issues/411

Cross-referencing #440 which is adding a `JSONObject` type, and in particular my comments at https://github.com/mozilla/uniffi-rs/pull/440#pullrequestreview-662789411 about whether we should be adding lots of builtin datatypes, or working on an extension...

Thanks for writing this up! Partly inspired by this issue, over in https://github.com/mozilla/uniffi-rs/pull/248 I've tried to write up some ADRs to explain the current set of tradeoffs that have lead...