Mateusz Burzyński

> I also noticed that the build is now failing, it doesn't really seem to be related to any of my changes unless I am missing something? It seems that...

> this means if you have a global machine but it isn't used in a component yet state transitions just get lost. Could you elaborate on what happens in this...

So I think what you are saying is that transitions won't get executed on the server side (because the machine is not started yet) and that can lead to a...

This has nothing to do with `assign`. With `setup` API our whole config is the target of inference - inferred type arguments are exempt from excess property check (which is...

You are not comparing the same flavor of `assign` here. If we want to compare the same flavors then we need to compare those two: ```ts assign({ foo: undefined, }),...

We can't address this as long as we allow property assigners derived from `Partial`. So the Bug label here isn't quite fitting (cc @davidkpiano). It feels like a design limitation...

@davidkpiano this likely should be removed from the docs for the time being since our types are actually not fully compatible with this flag, see https://github.com/statelyai/xstate/issues/4613

This works as expected according to the SCXML semantics of conflicting transitions (see [removeConflictingTransitions](https://www.w3.org/TR/scxml/#removeConflictingTransitions). However, we have tweaked rules around reentrancy of source states (see [here](https://github.com/statelyai/xstate/pull/4234/files#diff-c151c44da836faf40b7ce4be891b689de9bc37e383846104ac600c38e883946b)). Essentially, for many purposes...

You could auto-wrap those with https://github.com/4Catalyzer/babel-plugin-dev-expression

What kind of semantics you'd lose? I find `void` to be quite weird and problematic at times. Take a look at this: ```ts declare const test: string | void |...