Big Planification thread v0.5.0

[Champii]

• Proper trait and struct scoping

  • Need proper namespecing, i.e. stop ignoring IdentifierPath's NodeId
  • Resolve only traits that are in the current scope
  • Keep a higher-level all-in-one collection of traits, to help identify what to use

• Impl rework

  • Merge trait and struct impl into one generic method-collecting
  • Have a unified way to get a trait method and a struct method
    This implies that we need a real TraitSolver mechanism, like Chalk

• Type rework

  • Subtyping
    • Change State to have TypeId instead of Type
    • Have a map of TypeId=>Type inside the state
    • Have also a map of TypeId=>TypeId
    • And a map from TypeId=>TraitId for trait bounds
    • Add a solving phase where
      • We consider subtyping when resolving
      • Same for trait bounds
  • Trait bound
    • Change the syntax of traits to accept bounds
    • Add a check for every impl if the bounds are respected
  • Optional signature for fns
    • Reimplement optional type-signature to force a type for a fn #155
    • Have a pre-check of a sig to know if they need to have the stack-based constraint
      mechanism applied to it (in case of a generic or bound type somewhere) or if
      they can be infered right away and put out of this aforementioned mechanism
  • Signature bound
    • Add a bound for trait applied to generics at the begining of the sig
    • When constraining, add bounds to TypeIds that propagate when solving

• Struct generics

• Enums (and generics)

This thread's purpose is to create, organize and track issues realated to the work needed to
release v0.5.0. For this release we aim to have a real proper type system with a real proper inference
system.

The checklist above is meant to be edited at any time to add, amend or complete listed tasks.
Sections that have a good enough planification will be transformed into issues for dispatch and development.

Feel free to pop in to comment and propose any improvement.