// Copyright 2020 The Go Authors. All rights reserved. // Use of this source code is governed by a BSD-style // license that can be found in the LICENSE file. // This file implements type unification. package types import ( "bytes" "go/token" "sort" ) // The unifier maintains two separate sets of type parameters x and y // which are used to resolve type parameters in the x and y arguments // provided to the unify call. For unidirectional unification, only // one of these sets (say x) is provided, and then type parameters are // only resolved for the x argument passed to unify, not the y argument // (even if that also contains possibly the same type parameters). This // is crucial to infer the type parameters of self-recursive calls: // // func f[P any](a P) { f(a) } // // For the call f(a) we want to infer that the type argument for P is P. // During unification, the parameter type P must be resolved to the type // parameter P ("x" side), but the argument type P must be left alone so // that unification resolves the type parameter P to P. // // For bidirection unification, both sets are provided. This enables // unification to go from argument to parameter type and vice versa. // For constraint type inference, we use bidirectional unification // where both the x and y type parameters are identical. This is done // by setting up one of them (using init) and then assigning its value // to the other. // A unifier maintains the current type parameters for x and y // and the respective types inferred for each type parameter. // A unifier is created by calling newUnifier. type unifier struct { check *Checker exact bool x, y tparamsList // x and y must initialized via tparamsList.init types []Type // inferred types, shared by x and y } // newUnifier returns a new unifier. // If exact is set, unification requires unified types to match // exactly. If exact is not set, a named type's underlying type // is considered if unification would fail otherwise, and the // direction of channels is ignored. func newUnifier(check *Checker, exact bool) *unifier { u := &unifier{check: check, exact: exact} u.x.unifier = u u.y.unifier = u return u } // unify attempts to unify x and y and reports whether it succeeded. func (u *unifier) unify(x, y Type) bool { return u.nify(x, y, nil) } // A tparamsList describes a list of type parameters and the types inferred for them. type tparamsList struct { unifier *unifier tparams []*TypeName // For each tparams element, there is a corresponding type slot index in indices. // index < 0: unifier.types[-index-1] == nil // index == 0: no type slot allocated yet // index > 0: unifier.types[index-1] == typ // Joined tparams elements share the same type slot and thus have the same index. // By using a negative index for nil types we don't need to check unifier.types // to see if we have a type or not. indices []int // len(d.indices) == len(d.tparams) } // String returns a string representation for a tparamsList. For debugging. func (d *tparamsList) String() string { var buf bytes.Buffer buf.WriteByte('[') for i, tname := range d.tparams { if i > 0 { buf.WriteString(", ") } writeType(&buf, tname.typ, nil, nil) buf.WriteString(": ") writeType(&buf, d.at(i), nil, nil) } buf.WriteByte(']') return buf.String() } // init initializes d with the given type parameters. // The type parameters must be in the order in which they appear in their declaration // (this ensures that the tparams indices match the respective type parameter index). func (d *tparamsList) init(tparams []*TypeName) { if len(tparams) == 0 { return } if debug { for i, tpar := range tparams { assert(i == tpar.typ.(*_TypeParam).index) } } d.tparams = tparams d.indices = make([]int, len(tparams)) } // join unifies the i'th type parameter of x with the j'th type parameter of y. // If both type parameters already have a type associated with them and they are // not joined, join fails and return false. func (u *unifier) join(i, j int) bool { ti := u.x.indices[i] tj := u.y.indices[j] switch { case ti == 0 && tj == 0: // Neither type parameter has a type slot associated with them. // Allocate a new joined nil type slot (negative index). u.types = append(u.types, nil) u.x.indices[i] = -len(u.types) u.y.indices[j] = -len(u.types) case ti == 0: // The type parameter for x has no type slot yet. Use slot of y. u.x.indices[i] = tj case tj == 0: // The type parameter for y has no type slot yet. Use slot of x. u.y.indices[j] = ti // Both type parameters have a slot: ti != 0 && tj != 0. case ti == tj: // Both type parameters already share the same slot. Nothing to do. break case ti > 0 && tj > 0: // Both type parameters have (possibly different) inferred types. Cannot join. return false case ti > 0: // Only the type parameter for x has an inferred type. Use x slot for y. u.y.setIndex(j, ti) default: // Either the type parameter for y has an inferred type, or neither type // parameter has an inferred type. In either case, use y slot for x. u.x.setIndex(i, tj) } return true } // If typ is a type parameter of d, index returns the type parameter index. // Otherwise, the result is < 0. func (d *tparamsList) index(typ Type) int { if t, ok := typ.(*_TypeParam); ok { if i := t.index; i < len(d.tparams) && d.tparams[i].typ == t { return i } } return -1 } // setIndex sets the type slot index for the i'th type parameter // (and all its joined parameters) to tj. The type parameter // must have a (possibly nil) type slot associated with it. func (d *tparamsList) setIndex(i, tj int) { ti := d.indices[i] assert(ti != 0 && tj != 0) for k, tk := range d.indices { if tk == ti { d.indices[k] = tj } } } // at returns the type set for the i'th type parameter; or nil. func (d *tparamsList) at(i int) Type { if ti := d.indices[i]; ti > 0 { return d.unifier.types[ti-1] } return nil } // set sets the type typ for the i'th type parameter; // typ must not be nil and it must not have been set before. func (d *tparamsList) set(i int, typ Type) { assert(typ != nil) u := d.unifier switch ti := d.indices[i]; { case ti < 0: u.types[-ti-1] = typ d.setIndex(i, -ti) case ti == 0: u.types = append(u.types, typ) d.indices[i] = len(u.types) default: panic("type already set") } } // types returns the list of inferred types (via unification) for the type parameters // described by d, and an index. If all types were inferred, the returned index is < 0. // Otherwise, it is the index of the first type parameter which couldn't be inferred; // i.e., for which list[index] is nil. func (d *tparamsList) types() (list []Type, index int) { list = make([]Type, len(d.tparams)) index = -1 for i := range d.tparams { t := d.at(i) list[i] = t if index < 0 && t == nil { index = i } } return } func (u *unifier) nifyEq(x, y Type, p *ifacePair) bool { return x == y || u.nify(x, y, p) } // nify implements the core unification algorithm which is an // adapted version of Checker.identical0. For changes to that // code the corresponding changes should be made here. // Must not be called directly from outside the unifier. func (u *unifier) nify(x, y Type, p *ifacePair) bool { // types must be expanded for comparison x = expand(x) y = expand(y) if !u.exact { // If exact unification is known to fail because we attempt to // match a type name against an unnamed type literal, consider // the underlying type of the named type. // (Subtle: We use isNamed to include any type with a name (incl. // basic types and type parameters. We use asNamed() because we only // want *Named types.) switch { case !isNamed(x) && y != nil && asNamed(y) != nil: return u.nify(x, under(y), p) case x != nil && asNamed(x) != nil && !isNamed(y): return u.nify(under(x), y, p) } } // Cases where at least one of x or y is a type parameter. switch i, j := u.x.index(x), u.y.index(y); { case i >= 0 && j >= 0: // both x and y are type parameters if u.join(i, j) { return true } // both x and y have an inferred type - they must match return u.nifyEq(u.x.at(i), u.y.at(j), p) case i >= 0: // x is a type parameter, y is not if tx := u.x.at(i); tx != nil { return u.nifyEq(tx, y, p) } // otherwise, infer type from y u.x.set(i, y) return true case j >= 0: // y is a type parameter, x is not if ty := u.y.at(j); ty != nil { return u.nifyEq(x, ty, p) } // otherwise, infer type from x u.y.set(j, x) return true } // For type unification, do not shortcut (x == y) for identical // types. Instead keep comparing them element-wise to unify the // matching (and equal type parameter types). A simple test case // where this matters is: func f[P any](a P) { f(a) } . switch x := x.(type) { case *Basic: // Basic types are singletons except for the rune and byte // aliases, thus we cannot solely rely on the x == y check // above. See also comment in TypeName.IsAlias. if y, ok := y.(*Basic); ok { return x.kind == y.kind } case *Array: // Two array types are identical if they have identical element types // and the same array length. if y, ok := y.(*Array); ok { // If one or both array lengths are unknown (< 0) due to some error, // assume they are the same to avoid spurious follow-on errors. return (x.len < 0 || y.len < 0 || x.len == y.len) && u.nify(x.elem, y.elem, p) } case *Slice: // Two slice types are identical if they have identical element types. if y, ok := y.(*Slice); ok { return u.nify(x.elem, y.elem, p) } case *Struct: // Two struct types are identical if they have the same sequence of fields, // and if corresponding fields have the same names, and identical types, // and identical tags. Two embedded fields are considered to have the same // name. Lower-case field names from different packages are always different. if y, ok := y.(*Struct); ok { if x.NumFields() == y.NumFields() { for i, f := range x.fields { g := y.fields[i] if f.embedded != g.embedded || x.Tag(i) != y.Tag(i) || !f.sameId(g.pkg, g.name) || !u.nify(f.typ, g.typ, p) { return false } } return true } } case *Pointer: // Two pointer types are identical if they have identical base types. if y, ok := y.(*Pointer); ok { return u.nify(x.base, y.base, p) } case *Tuple: // Two tuples types are identical if they have the same number of elements // and corresponding elements have identical types. if y, ok := y.(*Tuple); ok { if x.Len() == y.Len() { if x != nil { for i, v := range x.vars { w := y.vars[i] if !u.nify(v.typ, w.typ, p) { return false } } } return true } } case *Signature: // Two function types are identical if they have the same number of parameters // and result values, corresponding parameter and result types are identical, // and either both functions are variadic or neither is. Parameter and result // names are not required to match. // TODO(gri) handle type parameters or document why we can ignore them. if y, ok := y.(*Signature); ok { return x.variadic == y.variadic && u.nify(x.params, y.params, p) && u.nify(x.results, y.results, p) } case *_Sum: // This should not happen with the current internal use of sum types. panic("type inference across sum types not implemented") case *Interface: // Two interface types are identical if they have the same set of methods with // the same names and identical function types. Lower-case method names from // different packages are always different. The order of the methods is irrelevant. if y, ok := y.(*Interface); ok { // If identical0 is called (indirectly) via an external API entry point // (such as Identical, IdenticalIgnoreTags, etc.), check is nil. But in // that case, interfaces are expected to be complete and lazy completion // here is not needed. if u.check != nil { u.check.completeInterface(token.NoPos, x) u.check.completeInterface(token.NoPos, y) } a := x.allMethods b := y.allMethods if len(a) == len(b) { // Interface types are the only types where cycles can occur // that are not "terminated" via named types; and such cycles // can only be created via method parameter types that are // anonymous interfaces (directly or indirectly) embedding // the current interface. Example: // // type T interface { // m() interface{T} // } // // If two such (differently named) interfaces are compared, // endless recursion occurs if the cycle is not detected. // // If x and y were compared before, they must be equal // (if they were not, the recursion would have stopped); // search the ifacePair stack for the same pair. // // This is a quadratic algorithm, but in practice these stacks // are extremely short (bounded by the nesting depth of interface // type declarations that recur via parameter types, an extremely // rare occurrence). An alternative implementation might use a // "visited" map, but that is probably less efficient overall. q := &ifacePair{x, y, p} for p != nil { if p.identical(q) { return true // same pair was compared before } p = p.prev } if debug { assert(sort.IsSorted(byUniqueMethodName(a))) assert(sort.IsSorted(byUniqueMethodName(b))) } for i, f := range a { g := b[i] if f.Id() != g.Id() || !u.nify(f.typ, g.typ, q) { return false } } return true } } case *Map: // Two map types are identical if they have identical key and value types. if y, ok := y.(*Map); ok { return u.nify(x.key, y.key, p) && u.nify(x.elem, y.elem, p) } case *Chan: // Two channel types are identical if they have identical value types. if y, ok := y.(*Chan); ok { return (!u.exact || x.dir == y.dir) && u.nify(x.elem, y.elem, p) } case *Named: // Two named types are identical if their type names originate // in the same type declaration. // if y, ok := y.(*Named); ok { // return x.obj == y.obj // } if y, ok := y.(*Named); ok { // TODO(gri) This is not always correct: two types may have the same names // in the same package if one of them is nested in a function. // Extremely unlikely but we need an always correct solution. if x.obj.pkg == y.obj.pkg && x.obj.name == y.obj.name { assert(len(x.targs) == len(y.targs)) for i, x := range x.targs { if !u.nify(x, y.targs[i], p) { return false } } return true } } case *_TypeParam: // Two type parameters (which are not part of the type parameters of the // enclosing type as those are handled in the beginning of this function) // are identical if they originate in the same declaration. return x == y // case *instance: // unreachable since types are expanded case nil: // avoid a crash in case of nil type default: u.check.dump("### u.nify(%s, %s), u.x.tparams = %s", x, y, u.x.tparams) unreachable() } return false }