Changes between Version 17 and Version 18 of TypeFunctionsTypeChecking
- Aug 16, 2006 6:15:24 PM (9 years ago)
v17 v18 19 19 Instances of indexed types are type checked by `TcTyClDecls.tcIdxTyInstDecl`; i.e., the same functions that performs their kind checking. Kind checking and type checking of instances of indexed types can be combined, as we don't need to worry as much about recursive dependencies as we have to for standard type declarations. In particular, the kinds of indexed types are declared by their signature and we don't have to compute any recursiveness information, as we never know whether we reached a fixed point for open types. (Hence, we conservatively assume indexed types to be always `Recursive`. This is safe as they are implicit loop breakers due to to implying coercions.) 20 20 21 == Representation of indexed families after type checking== 21 == == 22 22 23 23 The function `TcTyClsDecls.tcTyClsDecls` produces `TypeRep.TyThing`s from type and class declarations. The `TyThing`s produced from the new declaration forms are the following: … … 36 36 Consequently, all functions that dependent on this field need to be extended. In particular, `TcType.isTauTyCon` regards applications of type family constructors as ''tau types'', which is why we need to require that the right hand side of each `type instance` declaration is also a tau type. As a result, `BuildTyCls.buildSynTyCon`'s last argument now also takes a value of type `SynTyConRhs`. 37 37 38 39 40 41 38 42 === GHC API === 39 43 40 44 The GHC API has a new predicate `isOpenTyCon` with the understanding that it is illegal to invoke `synTyConDefn`, `synTyConRhs`, and `tyConDataCons` on type constructors that fulfil `isOpenTyCon`. 41 45 42 == Representation of type equation axioms == 46 == Core representation of instances of indexed types == 47 48 === Representation of type equation axioms === 43 49 44 50 Type functions have a number of properties in common with class instances; in particular, they require a machinery for matching type patterns against types, as instance heads do during context simplification. Hence, we probably want some structure similar to `InstEnv.Instance` for type functions - for instances this is maintained in the field `iSpec` of `TcEnv.InstInfo` (for type functions we don't need anything like `iBinds` as they are pure type-level entities). If possible, it would be ideal if we can reuse (or generalise) some of the matching machinery for instance heads.