Welcome to OStack Knowledge Sharing Community for programmer and developer-Open, Learning and Share
Welcome To Ask or Share your Answers For Others

Categories

0 votes
647 views
in Technique[技术] by (71.8m points)

multiple inheritance - Method resolution order in C++

Consider the following class hierarchy:

  • base class Object with a virtual method foo()
  • an arbitrary hierarchy with multiple inheritance (virtual and non-virtual); each class is a subtype of Object; some of them override foo(), some don't
  • a class X from this hierarchy, not overriding foo()

How to determine which method will be executed upon a call of foo() on an object of class X in C++?

(I'm looking for the algorithm, not any specific case.)

See Question&Answers more detail:os

与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…
Welcome To Ask or Share your Answers For Others

1 Answer

0 votes
by (71.8m points)

There is no MRO in C++ like Python. If a method is ambiguous, it is a compile-time error. Whether a method is virtual or not doesn't affect it, but virtual inheritance will.


The algorithm is described in the C++ standard §[class.member.lookup] (10.2). Basically it will find the closest unambiguous implementation in the superclass graph. The algorithm works like this:

  1. Suppose you want to look up a function f in class C.

  2. We define a look-up set S(f, C) being a pair of sets (Δ, Σ) representing all possibilities. (§10.2/3)

    • The set Δ is called the declaration set, which is basically all the possible f's.

    • The set Σ is called the subobject set, which contain the classes that these f's are found.

  3. Let S(f, C) include all f directly defined (or using-ed) in C, if any (§10.2/4):

    Δ = {f in C};
    if (Δ != empty)
      Σ = {C};
    else
      Σ = empty;
    S(f, C) = (Δ, Σ);
    
  4. If S(f, C) is empty (§10.2/5),

    • Compute S(f, Bi) where Bi is a base class of C, for all i.

    • Merge each S(f, Bi) into S(f, C) one by one.

      if (S(f, C) == (empty, empty)) {
        B = base classes of C;
        for (Bi in B)
          S(f, C) = S(f, C) .Merge. S(f, Bi);
      }
      
  5. Finally the declaration set is returned as the result of name resolution (§10.2/7).

    return S(f, C).Δ;
    
  6. The merge between two look-up sets (Δ1, Σ1) and (Δ2, Σ2) is defined as (§10.2/6):

    • If every class in Σ1 is a base class of at least one class in Σ2, return (Δ2, Σ2).
      (Similar for the reverse.)
    • Else if Δ1Δ2, return (ambiguous, Σ1Σ2).
    • Otherwise, return (Δ1, Σ1Σ2)

      function Merge ( (Δ1, Σ1), (Δ2, Σ2) ) {
      
         function IsBaseOf(Σp, Σq) {
           for (B1 in Σp) {
             if (not any(B1 is base of C for (C in Σq)))
               return false;
           }
           return true;
         }
      
         if      (Σ1 .IsBaseOf. Σ2) return (Δ2, Σ2);
         else if (Σ2 .IsBaseOf. Σ1) return (Δ1, Σ1);
         else {
            Σ = Σ1 union Σ2;
            if (Δ1 != Δ2)
              Δ = ambiguous; 
            else
              Δ = Δ1;
            return (Δ, Σ);
         }
      }
      

For example (§10.2/10),

struct V { int f(); };
struct W { int g(); };
struct B : W, virtual V { int f(); int g(); };
struct C : W, virtual V { };

struct D : B, C {
   void glorp () {
     f();
     g();
   }
};

We compute that

S(f, D) = S(f, B from D) .Merge. S(f, C from D)
        = ({B::f}, {B from D}) .Merge. S(f, W from C from D) .Merge. S(f, V)
        = ({B::f}, {B from D}) .Merge. empty .Merge. ({V::f}, {V})
        = ({B::f}, {B from D})   // fine, V is a base class of B.

and

S(g, D) = S(g, B from D) .Merge. S(g, C from D)
        = ({B::g}, {B from D}) .Merge. S(g, W from C from D) .Merge. S(g, V)
        = ({B::g}, {B from D}) .Merge. ({W::g}, {W from C from D}) .Merge. empty
        = (ambiguous, {B from D, W from C from D})  // the W from C is unrelated to B.

与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…
Welcome to OStack Knowledge Sharing Community for programmer and developer-Open, Learning and Share
Click Here to Ask a Question

...