Guy Bedford
Guy Bedford
Nothing right now, mainly I just wanted a tracking issue to note that this is planned for when the specification is ready to be used. I suppose in theory we...
Following up on this one again - @nicolo-ribaudo is there anything we still need to get here, or do we get this integration naturally in the current spec model?
Very interesting approach. It would be great to see some tests for the sorts of workflows you're seeking to enable here, and that would also help for review too.
Yes, in the tests folder, we have one big test app that is built altogether and tested as one.
You should just be able to add a new binding there, if there's issues with that happy to look into it further.
> The idea behind that was to split out the different return types, for streaming replies we don't expect a type T that implements Deserialize/Serialize like we would normally do...
> No as there needs to be an additional wrapper as workers ai stream returns see events rather than just the T we need, so there needs to be a...
I'm still happy to review this for the release tomorrow if there is interest in finishing up the last change requests.
The concept of untyped forwarding is a difficult one to translate into Wasm in a way that is meaningful. Even with type enforcement, bindings would go through reinterpretation that effectively...
I ended up exploring better reexport bindings in https://github.com/WebAssembly/esm-integration/pull/105 which only gain marginal benefits in very constructed cases with cycles. There is also strong resistance to deeper cyclical graph embedding,...