SasakiSaki

I understand, so is it possible to have instructions with semantics similar to the following: ```wat i32.const 1 ;; overrride defaultable table.idx funcref ;; fill undefaultable struct.new_default_with_field $type $field1 $field2...

As a temporary solution, you can try [Intellij WIT](https://plugins.jetbrains.com/plugin/24001-wit-idl), even the community version can use this plugin. It supports formatting the selected area, or the entire file, or the wit...

The current formatter is written in pure kotlin. In the long term, I may switch to a rust native implementation via the `document/formatter` protocol in lsp. If you want to...

Do I need to manually call `nn:buffer/g-tensor#[dtor]g-tensor` on the js side to ensure the release? If it has been released by gc, what will happen if I call it again....

Ok I need some time to confirm the actual behavior. This thing is difficult to debug because the debugger also holds a reference to it, which prevents recycling. I can...

Does wit-component have a concept similar to weak reference, so that it can hold its reference but not affect its life cycle?

I achieved the output html in my fork by changing the formatting here https://github.com/zesterer/ariadne/blob/ccd465160129d2546d67f7ee78727a4098a64330/src/draw.rs#L199 ```rust write!(f, "{}", col, Paint::new(&self.0).fg(col)) ``` --- But the original stderr write can't be used, here...

I thought of a mechanism like this to support different output formats: ```rust trait SpanFormatter { fn write_lhs(&self, w: impl Write) -> std::fmt::Result; fn write_rhs(&self, w: impl Write) -> std::fmt::Result;...

Yes, that's a good idea. The fallback scheme reminds me of the shader language. Faced with the huge difference in gpu capabilities, shaders have found a widely used scheme. If...

In this case, we need a mechanism to append information to the report.