bendk

> > we have to assume that the foreign code might have another reference to the object > But it has to be freed at some point, is this not...

> One question, more for @bendk and @gruberb: would you recommend @Standard8 start with UniFFI's proc macros (I think you're still working on the Gecko JS backend's support for them?),...

> We can use subfolders in the `docs` branch. Which means we have to integrate this workflow with the `generate-docs` workflow. We then create a subfolder there called `swift-docs`, which...

I'm hoping to merge this one today, so that we can have a real `mozilla/application-services` rev for the moz-central [patch](https://phabricator.services.mozilla.com/D219800) for this. I'll wait until https://github.com/mozilla/application-services/pull/6180 has been merged. That...

> I would suggest to align the philosophy of this client with the one we have on desktop 🙏 > > Instead of having a new method `get_cached_records()` etc., we...

> That would work for me, but as Mark said we have to clarify how we combine this with packaged dumps. For example, if a collection has a packaged dump,...

I'm going to close this one since we want to update this API to bring it closer to the one used by the official client and this code is now...

Okay, so let's merge this and https://github.com/mozilla/application-services/pull/6326, then move to the m-c work? I didn't quite understand the part about the nightly, but this seems like the path to me.

> a-s gets nightlies and that auto-opens a bug against m-c. This is the Android nightly right? I don't think there's something like this for Desktop. Does the Android nightly...

[The code](https://github.com/mozilla/uniffi-rs/blob/6d72caeccd008aafe70926ae3856dea64bbfd52e/uniffi_bindgen/src/bindings/python/templates/ObjectTemplate.py#L9) uses `has_callback_interface()` to check, which is only true for trait interfaces with `with_foreign`. That one confuses me every time. > isinstance probably doesn't matter as much as pyright...