Microsoft Word OpenIssuesInOO. doc

Class-based versus prototype-based languages. Prototype-based languages offer 

interesting alternatives to class-based languages with respect to modeling and technical 

possibilities. It has been argued that it should be possible to unite the two approaches. 

From a modeling point of view, the two alternatives seem to support different phases of 

the modeling activity. Technically there don't seem to be major differences in the sense 

that the new possibilities offered by prototype-based languages can be integrated with 

class- based languages. The  real difference seems to be with respect to philosophy: 

some people may prefer a pure class-based approach whereas others may prefer a pure 

prototype-based approach.  

• 

Concurrency. It has been argued that the Simula approach to active objects is still one 

of the best alternatives for supporting concurrency in object-oriented languages. 

There are a number of issues that have not been covered in this paper, but where there also 

seems to be some disagreement. Examples are: 

• 

Multi-methods. Some people argue that the unification of abstraction mechanisms into 

the concept of pattern as in BETA and the unification of class and object as in Self is 

going too far in the generalisation of programming language constructs. Multi-methods 

is another example of a generalization that may be unnecessary. They are definitely 

more general and more powerful than single-methods. Except for CLOS, multi-

methods have not been adapted in practice. To the present author it is open whether or 

not multi-methods are really needed, since the few practical examples encountered can 

be handled by double dispatch. 

•