Mark Hammond

huh - line 2 of that file is identical - so we are just repeating the record docstring for every field, which can be seen in the docstring test: ```...

@joe-p what do you think about this?

This is failing with: ``` def __init__(self, *, no_default_string:str, boolean:bool = _DEFAULT, integer:int = _DEFAULT, float_var:float = _DEFAULT, vec:typing.List[bool] = _DEFAULT, opt_vec:typing.Optional[typing.List[bool]] = _DEFAULT, opt_integer:typing.Optional[int] = _DEFAULT, custom_integer:CustomInteger = _DEFAULT,...

It's a placeholder - mainly to handle the "default value" case rather than null. We can't unconditionally use literals due to list etc. But even in the None case it's...

(I mean I guess we probably could use literals everywhere other than where they cause problems, but that still leaves lists, maps and objects)

> I mean I guess we probably could use literals everywhere other than where they cause problems That actually worked out really well and made the generated code for default...

Rust has no concept of a "main thread", so I think something like this will need to be restricted to the bindings. In our current world, this implies we might...

I believe we are unlikely to offer a capability where the Rust side of the world is only safe if the user promises the foreign binding doesn't do unsafe things,...

This does seem like a bug, but I can't repro it. eg, in our repo and the following patch: ``` --- a/fixtures/ext-types/uniffi-one/src/lib.rs +++ b/fixtures/ext-types/uniffi-one/src/lib.rs @@ -49,4 +49,9 @@ fn hello(&self)...

I don't think procmacros would make a difference here, but both those crates I mentioned above do have udl files.