How to use call_resumable together with consume_fuel?

[tomaka]

I'd like to call a Wasm function using the relatively new call_resumable feature, while at the same time use the Config::consume_fuel(true) feature. When the VM runs out of fuel, I would like the execution to be interrupted so that I can resume it later.

My question is: how do these two features work together?

I can see two options, and I don't know which one to expect.

• When an OutOfFuel trap happens, call_resumable returns a ResumableCall::Resumable where the host_error lets me know what kind of trap happened. In that situation, does that mean that every trap leads to ResumableCall::Resumable?
• When an OutOfFuel trap happens, call_resumable returns an error. However, in that situation, I don't know how to resume the execution if I simply want to re-fuel the store and ignore the trap.

Thanks for the help

[Robbepop]

As of now resumable calls can only be resumed for errors originating from called host functions. If an OutOfFuel trap is raised in Wasm code the trap is not resumable. If however, a called host functions manually consumes fuel via the Caller parameter that leads to an OutOfFuel trap then the OutOfFuel trap is originating from the host and therefore the call is resumable.

[tomaka]

Thanks for the answer

If I understand correctly, there's no way to do in wasmi something similar to this wasmtime behavior: https://docs.rs/wasmtime/latest/wasmtime/struct.Store.html#method.out_of_fuel_async_yield?

[Robbepop]

Correct. I was not aware of this configuration in Wasmtime. Maybe we can add something similar to wasmi. Although there is no notion of "async" calls in wasmi.

[Robbepop]

Please create an issue for wasmi with your constraints and usage scenario if you need this to be implemented.
The more info we get the better design we will find.

[Lucky4Luuk]

I'm looking for this exact feature. I know it's only been 4 days, but is there any progress on this? This seems to be the only reasonable choice right now for wasm in a no_std environment

[tomaka]

#696