[more comments and fixmes]

Author: fmease

compiler/rustc_hir_analysis/src/astconv/bounds.rs

@@ -401,7 +401,7 @@ impl<'tcx> dyn AstConv<'tcx> + '_ {
             let bound_vars = tcx.late_bound_vars(binding.hir_id);
             ty::Binder::bind_with_vars(subst_output, bound_vars)
         } else {
-            // Append the generic arguments of the associated type to the `trait_ref`.
+            // Append the generic arguments of the associated type or const to the `trait_ref`.
             candidate.map_bound(|trait_ref| {
                 let ident = Ident::new(assoc_item.name, binding.ident.span);
                 let item_segment = hir::PathSegment {
@@ -421,6 +421,9 @@ impl<'tcx> dyn AstConv<'tcx> + '_ {
 
                 debug!(?args_trait_ref_and_assoc_item);
 
+                // FIXME(fmease): Explain that we are indeed constructing an alias "type" for
+                // associated constants with feature `associated_const_equality` and that we
+                // call them "const projections".
                 ty::AliasTy::new(tcx, assoc_item.def_id, args_trait_ref_and_assoc_item)
             })
         };
@@ -437,6 +440,8 @@ impl<'tcx> dyn AstConv<'tcx> + '_ {
                     hir::Term::Const(ct) => ty::Const::from_anon_const(tcx, ct.def_id).into(),
                 };
 
+                // FIXME(fmease): Is this strictly necessary for `Term::Const`?
+                // If so, update the diagnostics, otherwise, skip this check.
                 if !speculative {
                     // Find any late-bound regions declared in `ty` that are not
                     // declared in the trait-ref or assoc_item. These are not well-formed.