Sadek Drobi’s Blog

in my previous post http://sadekdrobi.com/?p=25 i listed some  Satisfaction Confusions that we are having in

popular imperative procedural object oriented statically typed C-like syntax programming languages, the question was where the goodness really comes from, and what are the drawbacks to such confusions.

One epression that is lately often used to stegmatise a piece of code is to tell “not object oriented”.  This expression is a good sign of something we always waited, which is the awareness of the majority of today’s developers of  the importance of the Object Oriented programming principiles. But on the other hand, the expression itself highlights another satisfaction confusion. Even myself, i sometimes feel it an insult to be told that my code is not object oriented, so what?!! should it be? or… it is not meant to be !!!

 Yes it is not, consider microsoft’s dotNet’s delegates, they are methods holders, they are not object (forgetting about the implementation behind) ,  each time i talk about delegates (and spesifically about observer pattern implementation in dotNet) to a Java guru, i get an instant reaction  “hey! thats not oop” or “you cant compare that to the royal anonymous class” , and my answer is SURE YOU CAN NOT, java anonymous classes are a lot of goodness, but delegates are quite handy, why to construct an interface when u need to pass really one method.

The goodness of delegates makes observer’s usage in dotnet is quite easy , as easy as to subscribe to an event Event;

u do event+= a delegate that carries your method that has the suitable signature, and if your really think about it, you’ll find out that passing functions arround is not a microsoft’s inventery, that all what Functional Programming about…

consider the following example:

this piece of code enumerates a list of integers and returns a new list of their squares

IList newList= new List();

foreach(int i in list){

 newList.Add(i*i);

}

what if i want to get a list of every element i*2 , or maybe something else, i have each time to rewrite the Enumeration , and do the think,

so if we consider this piece of code that returns the sum of the elements in the list

int sum=0;

foreach(int i in list){

 sum+=i; }

what if i want the sum of the squeres? it makes sense to pass a function huh?

In fact functional programming ( and others like logical Prolog ) features are quite missing in most of today’s industrial popular languages. Even if -in my opinion- there construction is more logical and meaning revealing than instruction based languages.

But using these features doesnt mean that we forget objects, nor modelling and other practices we ever learnt. The point is TO FREE OUR MINDS!

But what about performance? well today’s prformance problem is not anymore like yesterdays, now we are adding more several processors to the machine than adding a more powerfull one. But in imperative languages u tell the machine exactly how to do, whereas in other paradigms (like functional programming ) you tell it what to do, and you leave the machine the freedom of how to do it! delegating some of the tasks of optemization to the machine than the developer.

Adding to all of that the fact that this might be more  Domain Driven Designs’s Ubiquitous Language friendly ( see Eric Evans DDD ).

Think freely, think in terms of what you need, and not what you like, use OOP when you need it, functional when it makes sense, predicates where it really fits.

What we really need is a level of maturity of developers when thay can decide how to do, not in terms of fashion, nor of feelings or habite, but in terms of need, then we can make a better choice i guess…

Doesnt all of that make sense?