Name annotations vs arbitrary identifiers

[AndrewScheidecker]

While I think this kind of metadata could be useful for stuff like the host bindings, I don't think it's a good solution to the arbitrary names issue. Consider:

(module
  (func $lambda (@name "λ") ...)
  ... (call $lambda) ...
)

What should this "disassemble" to? You would need to add a way to store both names for $lambda in the name section, and for producers to produce something useful for both names. Otherwise you're going to end up with something like this:

(module
  (func $0 (@name "λ") ...)
  ... (call $0) ...
)

I think it's simpler to extend the identifier syntax to support arbitrary strings (as discussed in the original issue):

(module
  (func $"λ" ...)
  ... (call $"λ") ...
)

[tlively]

The discussion about this idea ended up being held on #21, so this can probably be closed.

[rossberg]

Closing as duplicate of #21.