Eric walked through the process of installing the v2.0 release candidate runtime operator on Kubernetes using kind and helm, available for reference now in the wasmCloud docs. He demonstrated deploying a Wasm component and curling it to show it's running, highlighting the quick setup process.
Lucas discussed the new HTTP implementation in wasmCloud v2, emphasizing moving the HTTP processing closer to the runtime to improve performance. This approach avoids the extra network hop present in v1's wRPC-driven HTTP provider.
The goal is a solid HTTP implementation within wash that supports HTTP, gRPC, and HTTP/2, reducing the need for custom WASI plugins.
HTTP is now tightly coupled to the wasmtime context, allowing it to be enabled or disabled based on use cases.
HTTP can be integrated with the runtime by implementing a router for allowing or denying HTTP requests from components, OR by implementing a host handler for a full HTTP stack, allowing custom HTTP transports.
By default, all outbound HTTP requests are blocked unless explicitly enabled.
Discussion: Wasmtime, WASI P3, and Async Components
The support for HTTP/2 in Wasmtime is discussed, with the goal of making it transparent for users.
A design change in P3 components makes them incompatible with current Wasmtime versions. Wasmtime 40, expected in December, is the target for alignment across the ecosystem.
wasmCloud v2 will easily upgrade from P2 to P3 components, and the main benefit of P3 is more ergonomic async code within language bindings.
The full release with async components is targeted for late January.