 | Phlip wrote:
> JTK wrote:
>
> > That desn't even make sense Philip. C++ most certainly allows
> > "foo.bar()", whether bar() is declared virtual or not. Inheritance
or
> > lack therof doesn't matter. One of us is missing something
here....
>
> In C++, given struct A { void bar(); } and struct B { void bar(); },
there
> are only two ways to write a foo.bar() where foo could be of type A
or B.
> They are macros and templates. A template in C++ is essentially a
typesafe
> macro. Neither permit foo's type to vary at runtime.
>
> So...
>
> Dynamic Typing-contrast with Static Typing. Languages like Smalltalk,
> Python or Objective C bind method calls entirely at
run-time.
objective-c is not fully dynamic, for instance, you can't add or remove
fields at runtime.
>
> aObject.method(argument);
>
> That line will evaluate for any aObject that has a .method(argument)
to
> call. Think of it like the compiler just reads "method" as a text
string,
> and looks for that string in the given object's member list.
>
> Static Typing-languages like Java, Eiffel, or C++ bind the type of
> objects at compile time, and select the target of messages
at
> run time based only on types appearing within an object's
> inheritance graph.
>
> Object &aObject = getObject();
> aObject.method(argument);
>
> That statement requires aObject, at compile time, to refer only to
objects
> created by classes within the same inheritance graph as Object. At
runtime,
> calls to .method() dispatch to the correct type of aObject, but
unrelated
> objects with a .method() could not have compiled.
They *check* the binding at compile time, but at runtime you can change
it anyway :)
In glibc you can always use LD_PRELOAD to override functions, methods,
or any kind of exported symbols.
|
|
