//! # Lattice Variables //! //! This file contains generic code for operating on inference variables //! that are characterized by an upper- and lower-bound. The logic and //! reasoning is explained in detail in the large comment in `infer.rs`. //! //! The code in here is defined quite generically so that it can be //! applied both to type variables, which represent types being inferred, //! and fn variables, which represent function types being inferred. //! It may eventually be applied to their types as well, who knows. //! In some cases, the functions are also generic with respect to the //! operation on the lattice (GLB vs LUB). //! //! Although all the functions are generic, we generally write the //! comments in a way that is specific to type variables and the LUB //! operation. It's just easier that way. //! //! In general all of the functions are defined parametrically //! over a `LatticeValue`, which is a value defined with respect to //! a lattice. use super::type_variable::{TypeVariableOrigin, TypeVariableOriginKind}; use super::InferCtxt; use crate::traits::ObligationCause; use rustc_middle::ty::relate::{RelateResult, TypeRelation}; use rustc_middle::ty::TyVar; use rustc_middle::ty::{self, Ty}; pub trait LatticeDir<'f, 'tcx>: TypeRelation<'tcx> { fn infcx(&self) -> &'f InferCtxt<'f, 'tcx>; fn cause(&self) -> &ObligationCause<'tcx>; // Relates the type `v` to `a` and `b` such that `v` represents // the LUB/GLB of `a` and `b` as appropriate. // // Subtle hack: ordering *may* be significant here. This method // relates `v` to `a` first, which may help us to avoid unnecessary // type variable obligations. See caller for details. fn relate_bound(&mut self, v: Ty<'tcx>, a: Ty<'tcx>, b: Ty<'tcx>) -> RelateResult<'tcx, ()>; } pub fn super_lattice_tys<'a, 'tcx: 'a, L>( this: &mut L, a: Ty<'tcx>, b: Ty<'tcx>, ) -> RelateResult<'tcx, Ty<'tcx>> where L: LatticeDir<'a, 'tcx>, { debug!("{}.lattice_tys({:?}, {:?})", this.tag(), a, b); if a == b { return Ok(a); } let infcx = this.infcx(); let a = infcx.inner.borrow_mut().type_variables().replace_if_possible(a); let b = infcx.inner.borrow_mut().type_variables().replace_if_possible(b); match (a.kind(), b.kind()) { // If one side is known to be a variable and one is not, // create a variable (`v`) to represent the LUB. Make sure to // relate `v` to the non-type-variable first (by passing it // first to `relate_bound`). Otherwise, we would produce a // subtype obligation that must then be processed. // // Example: if the LHS is a type variable, and RHS is // `Box`, then we current compare `v` to the RHS first, // which will instantiate `v` with `Box`. Then when `v` // is compared to the LHS, we instantiate LHS with `Box`. // But if we did in reverse order, we would create a `v <: // LHS` (or vice versa) constraint and then instantiate // `v`. This would require further processing to achieve same // end-result; in partiular, this screws up some of the logic // in coercion, which expects LUB to figure out that the LHS // is (e.g.) `Box`. A more obvious solution might be to // iterate on the subtype obligations that are returned, but I // think this suffices. -nmatsakis (&ty::Infer(TyVar(..)), _) => { let v = infcx.next_ty_var(TypeVariableOrigin { kind: TypeVariableOriginKind::LatticeVariable, span: this.cause().span, }); this.relate_bound(v, b, a)?; Ok(v) } (_, &ty::Infer(TyVar(..))) => { let v = infcx.next_ty_var(TypeVariableOrigin { kind: TypeVariableOriginKind::LatticeVariable, span: this.cause().span, }); this.relate_bound(v, a, b)?; Ok(v) } _ => infcx.super_combine_tys(this, a, b), } }