Bridge call roundtrip

[lucksus]

Context

This is basic bridging. But no checks and no capability tokens. If the bridge is set in the config the caller can call any zome function of the callee.

hdk::call for bridge calls

Added instance handle to hdk::call(), which can also be hdk::THIS_INSTANCE to call other local zomes.

hdk::call("test-callee", "greeter", "public", "token", "hello", JsonString::from("{}"))

implement bridged call via Context::container_api

core/src/nucleus/ribosome/api/call.rs:

    let handler = container_api.write().unwrap();

    let id = ProcessUniqueId::new();
    let request = format!(
        r#"{{"jsonrpc": "2.0", "method": "{}", "params": {}, "id": "{}"}}"#,
        method, input.fn_args, id
    );

    let response = handler
        .handle_request_sync(&request)
        .ok_or("Bridge call failed".to_string())?;

Bridge roundtrip test in container_api crate

The test container config sets up a caller and a callee instance. The callee is created completely in the tests section in container.rs with WAT code. In there is a function "hello" that returns the string "Holo World" (set as ascii in WAT). The caller is the new WASM Rust project "test-bridge-caller" so I can use hdk::call. Test shows that "Holo World" arrives at the caller side.

[lucksus]

This handle is the "instance_handle" that the bridge caller has to provide to hdk::call to reference this bridged instance.

[lucksus]

"test-callee" is the handle set in the container config for this bridge. This actually has to be defined in the DNA but currently the container does not enforce these handles to match up (next PR). Then we will also have a way of defining the needed bridges in the define_zome! macro below. A bridge handle would then be like the declaration of binding: if declared in the zome as required, zome code can expect a call to such a bridge to work because the container would not load a config in which a matching bridge definition was missing..

[Connoropolous]

At this point, it kind of looks like hdk::call should accept a struct maybe, instead of so many params

[lucksus]

yep, though it will lose at least one again soon when the capability stuff is completely switched to tokens..

[zippy]

exactly. also the capability name is in the current call and that will go away too.

[Connoropolous]

At this point, it kind of looks like hdk::call should accept a struct maybe, instead of so many params

[zippy]

This is super great. Woot!

[zippy]

woot woot!!!

[zippy]

exactly. also the capability name is in the current call and that will go away too.