Re: SoA Vs OO
From: Mark Nicholls (Nicholls.Mark_at_mtvne.com)
Date: 06/04/04
- Next message: Mark Nicholls: "Re: SoA Vs OO"
- Previous message: Daniel T.: "Re: Active Objects"
- In reply to: Robert C. Martin: "Re: SoA Vs OO"
- Next in thread: Cristiano Sadun: "Re: SoA Vs OO"
- Reply: Cristiano Sadun: "Re: SoA Vs OO"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 4 Jun 2004 00:52:42 -0700
> >For the past decade and a half, the software industry mantra has been
> >OO. We have heard repeatedly that an object-oriented design is more
> >maintainable, more reusable and more resilient to changing
> >requirements. Now suddenly the air is abuzz with Service-oriented
> >Architecture. Object-orientation does not seem to be cool any more!
> >
> >When I take a closer look at SoA, it seems a lot closer to functional
> >design than object-oriented design. When I look at services, I tend
> >to look at them in functional terms AND NOT object terms. Whatever
> >happened to those oft-repeated protestations that functional
> >decomposition is bad and object-orientation is good?
> >
> >Am I missing something, or is the pendulum swinging back towards
> >functional design? Whatever happened to those neat concepts like
> >inheritance and polymorphism?
>
> Service Oriented Architecture is not a bad idea (although
> communicating packets through the http port is an immensely stupid
> idea). It is completely orthogonal to OO. SoA can (and almost
> certainly will) be implemented in OO languages using OO designs (so
> long as the programmers understand OO design). Moreover, services are
> by their very nature polymorphic, so SoA is in some sense an extension
> of OO.
>
> Functional decomposition is not a bad idea, nor was it ever a bad
> idea. Indeed, all object oriented designs are functionally
> decomposed. OO just adds polymorphism and encapsulation to that
> decomposition. There was a time, in the 90's, where it was
> fashionable to criticize designs as being "functionally decomposed"
> meaning anti-OO. However, the fashion was naive.
>
To a degree I agree with the original post...SOA is not OO and in a
sense is by definition not stateful OO. Experience seems to show that
statefull OO for inter application communication is in general a bad
idea, especially synchronous and non local communication.
We can argue the toss whether stateless objects are in fact just a
list of functions i.e. a functional API - to me they are, and we've
been there - or at least similar.
I would design both systems and the infrastructure in a standard OO
manner, but the architecture itself is effectively a functional one.
- Next message: Mark Nicholls: "Re: SoA Vs OO"
- Previous message: Daniel T.: "Re: Active Objects"
- In reply to: Robert C. Martin: "Re: SoA Vs OO"
- Next in thread: Cristiano Sadun: "Re: SoA Vs OO"
- Reply: Cristiano Sadun: "Re: SoA Vs OO"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
