🐞 [Bug]: method that requires state, but does not switch_to, returns stateless object.

[jkneer]

What happened?

#[type_state(
    states = (Unset, FullyConfigured),
    slots = (Unset)
)]
struct BobTheBuilder {}

#[impl_state]
impl BobTheBuilder {
  #[require(Unset)]
  #[switch_to(FullyConfigured)]
  pub fn init() -> BobTheBuilder {
    BobTheBuilder {}
  }
}

  #[require(FullyConfigured)]
  pub fn dosth(self, Option<u32>) -> BobTheBuilder {
    BobTheBuilder {}
  }

In this case the call to dosth() will return a stateless BobTheBuilder.
This is unexpected, because this basically switches the state from FullyConfigured to stateless without a switch_to statement. On the other hand this can be fixed by adding a switch_to macro to the initial state.
So this implementation

  #[require(FullyConfigured)]
  #[switch_to(FullyConfigured)]
  pub fn dosth(self, Option<u32>) -> BobTheBuilder {
    BobTheBuilder {}
  }

works as expected.

Expected behavior

This behaviour took me by surprise. I would have expected that the OutputState = InputState as long as I do not call switch_to and that I would not need to add switch_to as long as I just want to require a state and update the struct, but do not want to change the state. The current behaviour is kind of the inverse of what I expected by intuition.

[ozgunozerk]

You are absolutely right, if switch_to is not specified, it should not preserve the type.

Will deal with this soon, thanks for the report 💯

[ozgunozerk]

This one is resolved, will issue a new release after #35 is also resolved