Why does Delphi allow constructor parameters to be incorrect?

Question

This might seem like a really silly question, but I don't know why this is even allowed to compile:

program ConstructorWithParam;

{$APPTYPE CONSOLE}

uses
  System.SysUtils;

type

  TThing = class(TObject)
  private
    FParent: TObject;
  public
    constructor Create(const AParent: TObject);
  end;

{ TThing }

constructor TThing.Create; // <- WTF? Why does the compiler not complain?
begin
  FParent := AParent;
end;

var
  Thing: TThing;
begin
  try
    Thing := TThing.Create(TObject.Create);
    Readln;
  except
    on E: Exception do
      Writeln(E.ClassName, ': ', E.Message);
  end;
end.

I'm using Delphi XE5 and have not tested on other versions. Thanks.

Answer 1

The first declaration in the form class is presumed to be the correct one. The implementation version does not need to identify the parameters required; they're assumed by the original declaration. This is a part of the language itself.

Here's a good example to illustrate the point:

type
  TMyClass = class (Tobject)
    procedure DoSometimg(DoA, DoB: Boolean);
  end;

Implementation:

procedure TMyClass.DoSomething;   // Note both parameters missing
begin
  if DoA then    // Note not mentioned in implementation declaration
    DoOneThing;   // but still can be used here
  if DoB then
    DoAnotherThing;
end;

I personally prefer to make both the implementation and interface declarations match, because it makes it easier to identify the parameters without jumping around as much in the code editor.