How to run

kit run-tests

If you change any process, make sure to run hyper-bindgen (install at hyper-bindgen), then kit b.

Todos for the future

• Http stubs aren't fully generated yet. Using hyperware process lib to send_request (POST) should be fairly straightforward. Extending hyper-bindgens caller util generator with an example of how a client request works should be able to one shot it. This will allows using autogenerated rpc functions for rpc calls.
• Add support for LazyLoadBlob in a message body. This will allow sending larger messages within this framework.
• Autogenerate function stubs from wit files in js/ts for frontend work. This is the exact same as how caller-utils is generated, except for js/ts. This will allow your frontend code to call hyperware processes as if they were just functions.
• Add better logging for tests. print_to_terminal only works on failure, and logging could be very useful here to understand what is generally going on.
• Occasionally, even when having made changes to the code, kit b or kit run-tests doesn't trigger compilation, and one has to edit the files once more. This could cause problems during development with LLMs and is likely a kit issue.
• We currently use records in the wit files to represent function signatures. It is entirely possible to change these to regular functions that return a future. Future support for implementing a wit component will only be available with WASI preview 3, but one can change the stub generation code to already accomodate this. Procrastinating on this one until preview 3 is probably fine.
• Heavy dogfooding and performance tests.
• Test websocket client from a frontend.