Lukas Reining

> So you have blocking or non-blocking init (maybe by just setting a short timeout), and then you have a means to listen to events. In which case you have...

> Everything can be made to work. So I have no examples where it doesn't work. My problem is with the level of complexity required. If generally this is considered...

> In a concurrent language you can check a state and register a handler and there is no race condition as long as you do not await as a micro-task...

> The option allows it to be either fully controlled by the provider or fully controlled by the OF SDK (for simple providers). If it was always fully controlled by...

> Do you know why Flipt needs to cast flag types? @beeme1mr I think even though we have specified this in 1.4.1 [[1]](https://github.com/open-feature/spec/blob/main/specification/sections/01-flag-evaluation.md#requirement-134), we have, with Flipt and Flagsmith, at...

In #1157 we discussed (https://github.com/open-feature/js-sdk/pull/1157#pullrequestreview-2649047308) that our approach of using the `OpenFeatureApi` singleton for setting providers in a Next.js App can create race conditions or even security issues. In Next.js...

> I see your point, and if we were just writing an internal TS library I would be OK with that. But don't we want to support JS as well?...

I think you found a problem in the ADR there @erka. In my opinion, this is just a proposal for all those who want to optionally have the option to...

Mh I get that this is quite challenging and know the cases of enterprises with hundreds of flags @erka. But this is what really many of the provider SDKs are...

From my point of view, this is ready for release candidate!