During some workaround with some architecture I faced a question and I am wondering if this implementation follows best practice or not.
Here we have a base abstract class:
abstract class A {
protected mixed $object;
public function __construct() {
$this->loadObject()
->fuc1();
}
abstract protected function loadObject(): self;
abstract public function fuc1(): bool;
abstract public function fuc3(): iterable;
}
In this case, I want to implement Class B and C. if I create each class separately everything is fine but as far as these 2 classes have the same functions I decided to create a base class for them (BC
) so codes are now like this:
class BC extends A {
protected string x;
protected string y;
protected function loadObject(): self {
throw new \Exception('Child must implement loeadObject');
}
public function fuc1(): bool {
//Some Codes
}
public function fuc3(): iterable {
//Some Codes
}
}
And Class B and Class C are like this:
class B extends BC {
protected function loadObject(): self {
$this->object = new SomeObject();
return $this;
}
}
class C extends BC {
protected function loadObject(): self {
$this->object = new SomeObject2();
return $this;
}
}
We also can not move loadObject
function to class BC cuz maybe class D wants to inherit directly from class A.
We can also rely on interfaces but I was wondering if some one forget to make class B/C implement an Interface then we will have problem in class A.
So is it a good practice that we throw exceptions in the class BC and force other developers to don't forget to overwrite the loadObject
function?
CodePudding user response:
This is not a good practice, if you doesn't need to implement the loadObject
method in the BC
class, you should keep it abstract.
abstract class BC extends A {
protected string $x;
protected string $y;
public function fuc1(): bool {
}
public function fuc3(): iterable {
}
}