structs, classes, interfaces - Part II, Problem

[Martin Hinsch]

As has been explained by several people several times there is a good reasons 
for keeping structs and classes separate:
-  RTTI (in the general sense, includes vtables and such) makes classes
   byte-incompatible with the respective sequence of plain types. Classes with
   RTTI therefore can't be (easily) used as "masks" for plain sequences of
   bytes.
There are also good reasons to exclusively use reference semantics with 
classes:
- Slicing, that is the loss of information when assigning a subclass object to
  a superlass object is avoided.
- Consistent (runtime) polymorphism, i.e. since class objects are always 
  references method calls can be polymorphic per default (which would be 
  impossible in C++).

The (IMHO quite high) cost for solving these problems is the existence of two
relatively similar but largely incompatible concepts in the language.


-- 
Martin Hinsch

m.hinsch at rug.nl
+31 50 363 8097

CEES Centre for Ecological and Evolutionary Studies
Biologisch Centrum
Kerklaan 30
Postbus 14
9750 AA Haren