Preproposal: Shared cache and lifetime cycle at proc_macro context

[zzau13]

It is a poorly worked idea, but I think it could be put to good use.

• share a RwLock between proc_macros
  With the aim of being able to share intermediate representations and to be able to treat the nested ones.
• add lifetime synchronization cycle, like webpack or some other bundler
  In order to prioritize
• Unique identifier for types
  To be able to locate and solve

I think that of the types is currently not possible since they resolve after the proc_macro, but some approach can be tried

[burdges]

I'd suggest closing this issues and moving any discussion to https://users.rust-lang.org/ and maybe later https://github.com/rust-unofficial if it goes anywhere.

It's extremely messy handle lifetime-like constructs besides the actual lifetimes provided by rustc, even thread local storage sucks rust-random/rand#996 rust-random/rand#968 so really rust should require formal proofs of soundness for such things.

[zzau13]

Ok, I'm closing it and looking for another way more soundness