Peter Huene

I'm also saying you can't return an address into the memory for wasm-to-host, either. The adapter may free the string in the WebAssembly memory and the implementation of the free...

I'd also like to echo Alex's concerns with this mangling scheme. Per my limited understanding, these strings exist only to facilitate the "componentizer" (today `wit-component`, tomorrow `lld`?) by conveying the...

This is additionally compounded by the fact that mangled name strings are repeated (potentially many times) internally in components, both in instantiation argument names and in export aliasing. This mangling...

I should perhaps amend my statement that _this iteration_ of name mangling should not be included in the proposal. I don't think it's valuable to implement tooling around something we...

Honestly, that sounds much more complicated for the tooling than sticking everything in a custom section (or multiple) with an expressive format — one not inherently limited to core wasm...

Can you elaborate on what you think is missing from the binary format today that prevents us from synthesizing a wit definition from a component file? We have tooling that...

The tooling is expecting a unique type index for an aliased type in the wit, for example: ```wat (component (type $T (list string)) (alias outer 0 $T (type $A1)) (alias...

It seems to me that there's a myriad of problems of having full isomorphism with respect to `wit component types` as the former is name-based and the latter is index-based...

After discussing this more 1:1 with @pchickey, he pointed out a problem we'll have with the approach of "declaring reusable types as part of mangled name" around imports: the linker...

Hi @willemneal! So a WebAssembly component built with `cargo component` cannot be easily stripped to only the original core module without losing all of the information regarding how its imports...