Mark Hammond

> * I don't think there's anything stopping Kotlin from sharing a `PlacesConnection` between 2 threads I think the [mutex around our `PlacesDb`](https://github.com/mozilla/application-services/blob/a44feb6b6a2151bf094f3385145a0827a9fbc7cc/components/places/src/ffi.rs#L191) should prevent different threads concurrently using a...

In `.kt` and `.swift` files in this repo under components/places/android and components/places/ios

While I don't disagree that the pragmas should change, isn't a bigger problem that any SQL failure at startup is causing the app to crash?

I suspect it will end up depending on the specific component and the specific request, and I'm mildly against having our components manage this (ie, this would exist purely in...

> I'm mildly against having our components manage this uniffi decorators would possibly be a reasonable option :) But kotlin and swift changing to holding a nullable reference might be...

To make sure I understand, you'd like that param to actually be treated as an integer "depth" value rather than a boolean, right? (FWIW, it's not a boolean due to...

Looking at https://www.sqlite.org/limits.html#max_like_pattern_length, I suspect the problem is that the `query` param here is > 50k.

It looks like [HistoryMetadataSuggestionProvider](https://github.com/mozilla-mobile/android-components/blob/6ba529f9538b8ea9ffba1846821a76cc80c3de17/components/feature/awesomebar/src/main/java/mozilla/components/feature/awesomebar/provider/HistoryMetadataSuggestionProvider.kt#L54) takes the entire awesomebar contents and feeds it into this API. So isn't it possible that some extremely long URL is being edited? Places [handles URLs...

A while ago on Desktop, we saw some massive "normal" URLs, with insanely long params - eg, you could something like a data: URL, but encoded as a http url...

doh - `thread 'main' panicked at 'no support for external types yet'` - IIUC, @bendk added support for kotlin recently, but swift doesn't yet have support, so this can't possibly...