The one with virtual/dynamic
// declare in Child Class
constructor Create; virtual;
constructor TChildClass.Create;
begin
inherited;
end;
The one with override.
// declare in Child Class
constructor Create; override;
constructor TChildClass.Create;
begin
inherited;
end;
The one with nothing
// declare in Child Class
constructor Create;
constructor TChildClass.Create;
begin
inherited;
end;
Are these the same thing? It looks confusing.
Yes, there is a difference, but let's deal with the virtual keyword in more basic OOP terms first, yet still how it applies to Delphi methods.
When you declare a derived (child) class, and implement a method as "override", it means that you're overriding (surprise) the matching method of the base class.
This means that you can write code like this:
var child : TBaseClass;
begin
child := TChildClass.Create; // note that it is stored in TBaseClass variable
child.VirtualMethodDefinedInBaseClassThatHasBeenOverriddenInChildClass;
This will call the method in the child class, even if the variable is defined to be of the base class type. This is the whole purpose of virtual methods, you can access the object through a reference of a more general type, and still call methods that have been written for the particular type of object you're dealing with.
If you have a virtual method in the base class that you chose not to override in the child class, but instead reintroduce, you're effectively replacing it in some cases. Note that in most cases you need to tell the compiler that you really meant to do this, though I'm unsure about what Delphi requires here.
Basically, if your variable is of type TBaseClass, and you call a virtual method on it, that has been reintroduced in the TChildClass, it will still call the method in the base class.
However, if your variable is of type TChildClass, and you call that method on it, you will get the new method instead.
Now, for constructors, in Delphi, it is slightly different.
The point of virtual constructors is to be able to virtually construct objects, and to do that, Delphi also has "class types".
You can say this:
type TClassToUse = class of TBaseClass;
var cls : TClassToUse;
obj : TBaseClass;
begin
cls := TChildClass;
obj := cls.Create;
(note that my Delphi knowledge is a bit rusty here, if anyone spots bugs or glaring problems in the above code, please let me know or just fix it)
Here we store a "class" in a variable, and then ask the class to please construct an object for us. This allows us to switch out which class to create, but we also need to declare the constructors we want to use virtual, otherwise we will have problems.
So in the above code, if you declared the constructor as virtual in TBaseClass, and then override it in TChildClass (which the code is actually using in cls), the overridden constructor is the one that will be used.
If, on the other hand, you don't declare the constructor as virtual, we're back to the base class constructor. Virtual basically means to figure out the right method to execute at runtime, whereas non-virtual will figure it out at compile time.
Reintroduction as described for normal methods above, also works this way.
However, virtual constructors are only used as virtual when used through a class type.
No, static and virtual methods are not the same thing.
And override is a case of virtual method.
http://en.wikipedia.org/wiki/Virtual_function
http://docwiki.embarcadero.com/RADStudio/XE4/en/Methods#Method_Binding
Constructors bring nothing special here - they conform to the same rules as other methods for the question
Related
Is it ok to initialize members field before calling inherited in create ?
IE:
constructor TMyObject.create(AOwner: TComponent);
begin
fMyField := xxx;
inherited
end;
instead of normal way :
constructor TMyObject.create(AOwner: TComponent);
begin
inherited
fMyField := xxx;
end;
just to know is their is any drawback i didn't see ...
When an instance of a class is instantiated, the memory is allocated and default initialized (e.g. filled with zeros), and then the constructor is called. So, any code in a constructor executes after the default initialization, which would be the one timing issue that you might imagine scuppering what you are doing.
However, code such as yours is usually indicative of a deeper design malaise. How could it matter whether you initialized a value before calling the inherited constructor? There are two reasons I can imagine where you might be tempted to do this:
If the field in question is declared in your derived class, then the only way the ancestor code could access it is by calling a virtual (or dynamic) method. And doing so in a constructor is dangerous because the object is only partially created. That's a big toxic code smell.
If the field in question is declared in the ancestor classes, you might be using this mechanism to in effect pass an argument from derived class to ancestor. That's a rather weird way of doing it. A much more appropriate way would be to use arguments in your constructor.
Let's assume you have a very well designed Delphi project, which respects the dependency injection and some other good practices.
Now let's assume you need to mock a class defined as:
TMyClass = class
public
procedure Method1;
procedure Method2
end;
Method1 and Method2 are not virtual. What do you do in this case? To mock an object we need to inherit it and override every method you want to mock, but it is not possible in this case because they are not virtual. Should I change the source code to add virtual on every method I need to mock? Is not it bad?
Edit
I was thinking about creating a compiler directive to make all fields in the class to be virtual, is it a good ideia? Only my test suite is going to set the compiler directive.
EDIT2*
Embarcadero should provide an easy way of changing a method pointer of a class to another method point, without the need for virtual.
Make the methods virtual so you can mock them. (They don't need to be abstract.)
If you cannot do that, then wrap the class with another class. Make the wrapper's methods virtual, and in the default implementation, just forward call calls to the original class. Wherever your program uses the original class, replace it with the wrapper. Now, mock the wrapper.
To mock an object we need to inherit
it, but it is not possible in this
case.
I'd recommend that every class that you say "I need to mock this" should end up being based on an interface.
In other words, if you have methods that need to be mocked, they should be put into an interface, and then the class to be mocked implements that interface. Then, you create the mock by implementing the same interface in the mock object.
The alternative is to use a Mocking library. You can look at these SO question:
What is your favorite Delphi mocking library?
and this framework includes mocking objects as well:
http://code.google.com/p/emballo/
Mocking objects should also encourage the proper use of dependency injection in your code.
You should not only make the methods virtual, you should declare a pure virtual base class, and make the other classes use only the pure virtual base class name. Now you can use what we call the "Liskov substitution principle", and you can make as many concrete subtypes of the abstract base class as you need. Since the abstract base class works just like an Interface works in delphi, minus the reference counting part, you get the best of both worlds. You can really keep your app code simple, and reduce bad coupling this way, plus you get unit testable "composite objects" that you put together yourself.
// in UnitBase.pas
TMyClassBase = class
public
procedure Method1; virtual; abstract;
procedure Method2; virtual; abstract;
end;
// in UnitReal.pas
TMyClassReal = class(TMyClassbase)
public
procedure Method1; override;
procedure Method2; override;
end;
// in UnitMock.pas
TMyClassMock = class(TMyClassbase)
public
procedure Method1; override;
procedure Method2; override;
end;
In the place that uses "TMyClass", change it to use TMyClassbase:
TMyOtherClass = class(TMyOtherClassBase)
private
FMyThing:TMyClassbase;
public
property MyThing:TMyClassBase read FMyThing write FMyThing;
end;
By connecting TMyOtherclass at runtime you can decide whether to use the real or mock class:
// in my realapp.pas
MyOtherClassObj := TMyotherClass.Create;
MyOtherClassObj.MyThing := TMyOtherClassReal.Create; // real object
// in my unittest.pas
MyOtherClassObj := TMyotherClass.Create;
MyOtherClassObj.MyThing := TMyOtherClassMock.Create; // mock object
(Don't forget to make a way to free MyOtherClassObj.MyThing later, or you'll have a leak)
If I write
type
MyClass = class of TMyClass;
...
Obj := MyClass.Create;
the correct constructor (the one in TMyClass) is called.
If I write
var
ClassVar : TClass;
...
ClassVar := TMyClass;
Obj := ClassVar.Create;
only the TObject constructor is called.
Why? What's the difference between the two versions? Can I force the TMyClass constructor call in the second scenario?
TClass is declared as "Class of TObject" in system.pas. What constructor gets called is decided at compile-time, and all the compiler knows is what base class you're using. It doesn't know what the value of the variable will be when it runs, so it has to default to the base class. If you're using a TClass, then your base class is TObject.
If you're using a class variable, I assume you have some sort of heirarchy and you're trying to implement a factory. If you want to make sure the right constructor gets called based on the value of a class variable at runtime, not something contained in your code at compile time, you need a virtual constructor.
type
TMyBaseObject = class(TObject)
public
constructor Create; virtual;
end;
TMyClass = class of TMyBaseObject;
Use a TMyClass instead of a TClass as your class variable, and now the compiler will generate a call to TMyBaseObject.Create, which is virtual. Make sure all your derived classes override the base constructor, and you'll end up calling the right one at runtime.
TObject.Create is not virtual, you need to declare ClassVar as a classtype of a class with a virtual constructor.
I would suggest you look into overriding the AfterConstruction method that's introduced by the TObject to make polymorphism like this work.
Each class definition can introduce a new constructor, with it's own set of parameters. A class variable only knows of the constructor if the base class it's a variable for. This is all possible because a constructor is neither virtual nor overridden. You could flag the constructor virtual on your base-class, and flag all descending classes override, which blocks the parameter list. (I think if you forget 'override' the compiler warns your new constructir hides the virtual one.)
Or descend it from TComponent, which already have a virtual constructor.
I'm writing some software that targets two versions of very similar hardware which, until I use the API to initialize the hardware I'm not able to know which type I'll be getting back.
Because the hardware is very similar I planned to have a parent class (TParent) that has some abstract methods (for where the hardware differs) and then two child classes (TChildA, TChildB) which implement those methods in a hardware dependent manner.
So I would first instantiate an object of TParent check what kind it is then cast it to the correct child.
However when I do this and call one of the abstract methods fully implemented in the child class I get an EAbstractError.
e.g:
myHardware:=TParent.Create();
if myHardware.TypeA then
myHardware:=TChildA(myHardware)
else
myHardware:=TChildB(myHardware);
myHardware.SomeMehtod();
I'm assuming that I can't cast a Parent Class to a child class, and also that there's probably a better way of doing this. Any pointers?
You need a factory method to return you the correct class depending on the Type of hardware you are using...
function CreateHardware(isTypeA: Boolean): TParent;
begin
if IsTypeA then Result := TChildA.Create
else Result := TChildB.Create;
end;
...
var
myHardware: TParent;
begin
myHardware := CreateHardware(True);
myHardwarde.SomeMethod;
end;
... or you could use the State pattern.
Common in either approach is that your TParent class does not has the knowledge to determine the type of hardware.That knowlegde is transfered into the factory method, caller of the factory method, factory itself or state class.
Thanks to Binary Worrier and Mghie for pointing me in the right direction in this instance. The answer given by Lieven would be the easier way in cases where minimising the initialization of the hardware wasn't an issue.
The pImpl idiom is discussed elsewhere on SO
Here is how I understand the implementation in pseudo-delphi-code (note I've not bothered with public/private distinctions for this):
class TParent
procedure SomeMethod(); abstract;
end;
class TChildA (TParent)
procedure SomeMethod(); override;
end;
class TChildB (TParent)
procedure SomeMethod(); override;
end;
class THardware
HardwareSpecficMethods: TParent;
procedure SomeMethod;
constructor Create();
contrsuctor THardware.Create();
begin
InitializeHardware();
If Hardware.TypeA then
HardwareSpecificMethods:=TChildA.Create()
else
HardwareSpecificMethods:=TChildB.Create();
end;
procedure THardware.SomeMethod();
begin
HardwareSpecificMethods.SomeMethod();
end;
end; {class THardware}
You're right, you can't and shouldn't cast from base class to derived class.
I'm assuming you don't want to have the Child object re-run the Parent constructor?
If so . . .
Remove the Parent/Child relationship as it stands, you will have only one Hardware class.
For the specific ChildA and ChildB functionality, create a new inheritance pattern, so that you have an ISpecificHardwareTasks interface or base class, and two derived classes (SpecificA & SpecificB).
When Hardware is constructing it's self, and it gets to the point where it knows what type of hardware it's working with, it then creates an instance of SpecificA or SpecificB). This instance is private to Hardware.
Hardware exposes methods which wrap the ISpecificHardWareTasks methods (it can even implement that interface if that makes sense).
The Specific classes can take a reference to the Hardware class, if that's necessary (though I don't know if you have access to the this pointer in a constructor, my Delphi is getting rusty)
Hope these ramblings helped somewhat.
What was the motivation for having the reintroduce keyword in Delphi?
If you have a child class that contains a function with the same name as a virtual function in the parent class and it is not declared with the override modifier then it is a compile error. Adding the reintroduce modifier in such situations fixes the error, but I have never grasped the reasoning for the compile error.
If you declare a method in a descendant class that has the same name as a method in an ancestor class then you are hiding that ancestor method — meaning if you have an instance of that descendant class (that is referenced as that class) then you will not get the behavior of the ancestor. When the ancestor's method is virtual or dynamic, the compiler will give you a warning.
Now you have one of two choices to suppress that warning message:
Adding the keyword reintroduce just tells the compiler you know you are hiding that method and it suppresses the warning. You can still use the inherited keyword within your implementation of that descended method to call the ancestor method.
If the ancestor's method was virtual or dynamic then you can use override. It has the added behavior that if this descendant object is accessed through an expression of the ancestor type, then the call to that method will still be to the descendant method (which then may optionally call the ancestor through inherited).
So difference between override and reintroduce is in polymorphism. With reintroduce, if you cast the descendant object as the parent type, then call that method you will get the ancestor method, but if you access it the descendant type then you will get the behavior of the descendant. With override you always get the descendant. If the ancestor method was neither virtual nor dynamic, then reintroduce does not apply because that behavior is implicit. (Actually you could use a class helper, but we won't go there now.)
In spite of what Malach said, you can still call inherited in a reintroduced method, even if the parent was neither virtual nor dynamic.
Essentially reintroduce is just like override, but it works with non-dynamic and non-virtual methods, and it does not replace the behavior if the object instance is accessed via an expression of the ancestor type.
Further Explanation:
Reintroduce is a way of communicating intent to the compiler that you did not make an error. We override a method in an ancestor with the override keyword, but it requires that the ancestor method be virtual or dynamic, and that you want the behavior to change when the object is accessed as the ancestor class. Now enter reintroduce. It lets you tell the compiler that you did not accidentally create a method with the same name as a virtual or dynamic ancestor method (which would be annoying if the compiler didn't warn you about).
There are lots of answers here about why a compiler that lets you hide a member function silently is a bad idea. But no modern compiler silently hides member functions. Even in C++, where it's allowed to do so, there's always a warning about it, and that ought to be enough.
So why require "reintroduce"? The main reason is that this is the sort of bug that can actually appear by accident, when you're not looking at compiler warnings anymore. For example, let's say you're inheriting from TComponent, and the Delphi designers add a new virtual function to TComponent. The bad news is your derived component, which you wrote five years ago and distributed to others, already has a function with that name.
If the compiler just accepted that situation, some end user might recompile your component, ignore the warning. Strange things would happen, and you would get blamed. This requires them to explicitly accept that the function is not the same function.
The RTL uses reintroduce to hide inherited constructors. For example, TComponent has a constructor which takes one argument. But, TObject has a parameterless constructor. The RTL would like you to use only TComponent's one-argument constructor, and not the parameterless constructor inherited from TObject when instantiating a new TComponent. So it uses reintroduce to hide the inherited constructor. In this way, reintroduce is a little bit like declaring a parameterless constructor as private in C#.
First of all, "reintroduce" breaks the inheritance chain and should not be used, and I mean never ever. In my entire time I worked with Delphi (ca 10 years) I've stumbled upon a number of places that do use this keyword and it has always been a mistake in the design.
With that in mind here's the simplest way it works:
You have like a virtual method in a base class
Now you wanna have a method that has the exact same name, but maybe a different signature. So you write your method in the derived class with the same name and it will not compile because the contract is not fulfilled.
You put the reintroduce keyword in there and your base class does not know about your brand new implementation and you can use it only when accessing your object from a directly specified instance type. What that means is toy can't just assign the object to a variable of base type and call that method because it's not there with the broken contract.
Like I said it's pure evil and must be avoided at all cost (well, that's my opinion at least). It's like using goto - just a terrible style :D
The purpose of the reintroduce modifier is to prevent against a common logical error.
I will assume that it is common knowledge how the reintroduce keyword fixes the warning and will explain why the warning is generated and why the keyword is included in the language. Consider the delphi code below;
TParent = Class
Public
Procedure Procedure1(I : Integer); Virtual;
Procedure Procedure2(I : Integer);
Procedure Procedure3(I : Integer); Virtual;
End;
TChild = Class(TParent)
Public
Procedure Procedure1(I : Integer);
Procedure Procedure2(I : Integer);
Procedure Procedure3(I : Integer); Override;
Procedure Setup(I : Integer);
End;
procedure TParent.Procedure1(I: Integer);
begin
WriteLn('TParent.Procedure1');
end;
procedure TParent.Procedure2(I: Integer);
begin
WriteLn('TParent.Procedure2');
end;
procedure TChild.Procedure1(I: Integer);
begin
WriteLn('TChild.Procedure1');
end;
procedure TChild.Procedure2(I: Integer);
begin
WriteLn('TChild.Procedure2');
end;
procedure TChild.Setup(I : Integer);
begin
WriteLn('TChild.Setup');
end;
Procedure Test;
Var
Child : TChild;
Parent : TParent;
Begin
Child := TChild.Create;
Child.Procedure1(1); // outputs TChild.Procedure1
Child.Procedure2(1); // outputs TChild.Procedure2
Parent := Child;
Parent.Procedure1(1); // outputs TParent.Procedure1
Parent.Procedure2(1); // outputs TParent.Procedure2
End;
Given the above code both of the procedures in TParent are hidden. To say they are hidden means that the procedures can not be called through the TChild pointer. Compiling the code sample produces a single warning;
[DCC Warning] Project9.dpr(19): W1010 Method 'Procedure1' hides virtual method of base type 'TParent'
Why only a warning for the virtual function and not the other? Both are hidden.
A virtue of Delphi is that library designers are able to release new versions without fear of breaking the logic of existing client code. This contrasts to Java where adding new functions to a parent class in a library is fraught with danger because classes are implicitly virtual. Lets say that TParent from above lives in a 3rd party library, and the library manufacture releases the new version below.
// version 2.0
TParent = Class
Public
Procedure Procedure1(I : Integer); Virtual;
Procedure Procedure2(I : Integer);
Procedure Procedure3(I : Integer); Virtual;
Procedure Setup(I : Integer); Virtual;
End;
procedure TParent.Setup(I: Integer);
begin
// important code
end;
Imagine we had the following code in our client code
Procedure TestClient;
Var
Child : TChild;
Begin
Child := TChild.Create;
Child.Setup;
End;
For the client it does not matter if the code is compiled against version 2 or 1 of the library, in both cases TChild.Setup is called as the user intends. And in the library;
// library version 2.0
Procedure TestLibrary(Parent : TParent);
Begin
Parent.Setup;
End;
If TestLibrary is called with a TChild parameter, everything works as intended. The library designer have no knowledge of the TChild.Setup, and in Delphi this does not cause them any harm. The call above correctly resolves to TParent.Setup.
What would happen in a equivalent situation in Java? TestClient would work correctly as intended. TestLibrary would not. In Java all functions are assumed virtual. The Parent.Setup would resolve to TChild.Setup, but remember when TChild.Setup was written they had no knowledge of the future TParent.Setup, so they are certainly not going to ever call inherited. So if the library designer intended TParent.Setup to be called it will not be, no matter what they do. And certainly this could be catasrophic.
So the object model in Delphi requires explicit declaration of virtual functions down the chain of child classes. A side effect of this is that it is easy to forget to add the override modifier on child methods. The existence of the Reintroduce keyword is a convenience to the programmer. Delphi was designed so that the programmer is gently persuaded, by the generation of a warning, to explicitly state their intentions in such situations.
tl;dr: Trying to override a non-virtual method makes no sense. Add the keyword reintroduce to acknowledge that you're making a mistake.
Reintroduce tells the compiler you want to call the code defined in this method as an entry point for this class and its descendants, regardless of other methods with the same name in the ancestors’ chain.
Creating a TDescendant.MyMethod would create a potential confusion for the TDescendants in adding another method with the same name, which the compiler warns you about.
Reintroduce disambiguates that and tells the compiler you know which one to use.
ADescendant.MyMethod calls the TDescendant one, (ADescendant as TAncestor).MyMethod calls the TAncestor one. Always! No confusion…. Compiler happy!
This is true whether you want the descendant method to be virtual or not: in both cases you want to break the natural linkage of the virtual chain.
And it does not prevent you from calling the inherited code from within the new method.
TDescendant.MyMethod is virtual: ...but you cannot or don’t want to use the linkage.
You cannot because the method signature is different. You have no other choice as overriding is impossible in this case with return type or parameters not exactly the same.
You want to restart an inheritance tree from this class.
TDescendant.MyMethod is not virtual: You turn MyMethod into a static one at the TDescendant level and prevent further overriding. All classes inheriting from TDescendant will use the TDescendant implementation.
When the ancestor class also has a method with the same name, and it is not necessarily declared virtual, you would see a compiler warning (as you would hide this method).
In other words: You tell the compiler that you know that you hide the ancestor function and replace it with this new function and do so deliberately.
And why would you do this? If the method is virtual in the parent class, the only reason is to prevent polymorphism. Other then that just override and do not call inherited. But if the parent method is not declared virtual (and you cannot change that, because you do not own the code for example), you can inherit from that class and let people inherit from your class without seeing a compiler warning.
This has been introduced to the language because of Framework versions (including the VCL).
If you have an existing code base, and an update to a Framework (for instance because you bought a newer Delphi version) introduced a virtual method with the same name as a method in an ancestor of your code base, then reintroduce will allow you to get rid of the W1010 warning.
This is the only place where you should use reintroduce.
First, as it was said above, you should never ever deliberately reintroduce virtual method. The only sane use of reintroduce is when the author of the ancestor (not you) added a method that goes into conflict with your descendant and renaming your descendant method is not an option. Second, you can easily call the original version of the virtual method even in classes where you reintroduced it with different parameters:
type
tMyFooClass = class of tMyFoo;
tMyFoo = class
constructor Create; virtual;
end;
tMyFooDescendant = class(tMyFoo)
constructor Create(a: Integer); reintroduce;
end;
procedure .......
var
tmp: tMyFooClass;
begin
// Create tMyFooDescendant instance one way
tmp := tMyFooDescendant;
with tmp.Create do // please note no a: integer argument needed here
try
{ do something }
finally
free;
end;
// Create tMyFooDescendant instance the other way
with tMyFooDescendant.Create(20) do // a: integer argument IS needed here
try
{ do something }
finally
free;
end;
so what should be the purpose of reintroducing virtual method other than make things harder to read?
reintroduce allows you to declare a method with the same name as the ancestor, but with different parameters. It has nothing to do with bugs or mistakes!!!
For example, I often use it for constructors...
constructor Create (AOwner : TComponent; AParent : TComponent); reintroduce;
This allows me to create the internal classes in a cleaner fashion for complex controls such as toolbars or calendars. I normally have more parameters than that. Sometimes it is almost impossible or very messy to create a class without passing some parameters.
For visual controls, Application.Processmessages can get called after Create, which can be too late to use these parameters.
constructor TClassname.Create (AOwner : TComponent; AParent : TComponent);
begin
inherited Create (AOwner);
Parent := AParent;
..
end;