Stéphane Épardaud
Stéphane Épardaud
> I don't see any cases where we need to propagate DB Tx context through emitters. Well that's a very good question. What contexts need to be propagated through emitters?...
Could you share a minimal reproducer project please?
> Would it be possible to make a plan for it? Why would you want this?
Because MP-CP is widely supported in many projects that we depend on, works, and is not coming with a lot of things we don't want or need from J-Concurrency. We'd...
> > In fact, rather the opposite, it did not copy a key part of the spec we depend on > > Which part is that exactly? We're in the...
> But isn't this about MP/CP/A? That one seems less maintained (read; abandoned). Right? It's not seeing new features because it's a tiny mature spec with no need for new...
See, in Quarkus, we don't use that API, mostly. What we use is the rest of the API that deals with `ThreadContext` so we can apply contexts to Mutiny `Uni`...
In our experience, 100% of users of reactive in Quarkus use Mutiny, and I'm pretty sure a similar adoption rate wrt Reactor for frameworks based on that. This is due...
Context Propagation is mostly useful for Reactive use-cases. Blocking users rarely need context propagation because they can block their worker thread and not lose any context.
Long running actions should probably be reactive, so as not to pin down threads for a long time. We really discourage people from using managed executors in Quarkus.