www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - is "abstract" supposed to do something?

reply "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
in C++, there is a concept called an "abstract class", which is achieved by
having a class with at least one pure virtual function (function declaration
=0).  you are not allowed to create any instances of an abstract class.

i thought the keyword "abstract" would cause this behavior on a class
without the "implied" kind that you get with C++.  apparently not!

defining a class as "abstract" does not seem to do anything.  D allows you
to create instances of it.  deriving from the abstract class does not change
any behavior.

looking through the manual, i cannot find any explanation as to what the
abstract keyword does.

all i'm doing is this:

abstract class Parent
{
    int foo() { return 10; }
}

class Child : Parent
{
    int foo() { return 20; }
}

i'm still allowed to create instances of Parent and call its foo().

am i doing something wrong here?  am i not supposed to have any kind of
function declaration in the Parent class for it to be treated as abstract?
or does abstract do something completely different?  or is it even
implemented ;)
Jul 12 2004
next sibling parent reply Charlie <Charlie_member pathlink.com> writes:
It only works on methods, not entire classes.  For that i suggest an interface,
though it does seem to be either a flaw, or not documented well.

C

In article <ccuj6r$2k5p$1 digitaldaemon.com>, Jarrett Billingsley says...

in C++, there is a concept called an "abstract class", which is achieved by
having a class with at least one pure virtual function (function declaration
=0).  you are not allowed to create any instances of an abstract class.

i thought the keyword "abstract" would cause this behavior on a class
without the "implied" kind that you get with C++.  apparently not!

defining a class as "abstract" does not seem to do anything.  D allows you
to create instances of it.  deriving from the abstract class does not change
any behavior.

looking through the manual, i cannot find any explanation as to what the
abstract keyword does.

all i'm doing is this:

abstract class Parent
{
    int foo() { return 10; }
}

class Child : Parent
{
    int foo() { return 20; }
}

i'm still allowed to create instances of Parent and call its foo().

am i doing something wrong here?  am i not supposed to have any kind of
function declaration in the Parent class for it to be treated as abstract?
or does abstract do something completely different?  or is it even
implemented ;)
Jul 12 2004
next sibling parent Regan Heath <regan netwin.co.nz> writes:
On Mon, 12 Jul 2004 18:50:08 +0000 (UTC), Charlie 
<Charlie_member pathlink.com> wrote:


 It only works on methods, not entire classes.


Not true. See my example.


 For that i suggest an interface,


An interface does not help when you're trying to say, all derived classes 
*must* redefine this method *and* you want to provide a method body (so as 
derived classes can call super.method())


 though it does seem to be either a flaw, or not documented well.


Agreed.

Regan


 C

 In article <ccuj6r$2k5p$1 digitaldaemon.com>, Jarrett Billingsley says...

 in C++, there is a concept called an "abstract class", which is 
 achieved by
 having a class with at least one pure virtual function (function 
 declaration
 =0).  you are not allowed to create any instances of an abstract class.

 i thought the keyword "abstract" would cause this behavior on a class
 without the "implied" kind that you get with C++.  apparently not!

 defining a class as "abstract" does not seem to do anything.  D allows 
 you
 to create instances of it.  deriving from the abstract class does not 
 change
 any behavior.

 looking through the manual, i cannot find any explanation as to what the
 abstract keyword does.

 all i'm doing is this:

 abstract class Parent
 {
    int foo() { return 10; }
 }

 class Child : Parent
 {
    int foo() { return 20; }
 }

 i'm still allowed to create instances of Parent and call its foo().

 am i doing something wrong here?  am i not supposed to have any kind of
 function declaration in the Parent class for it to be treated as 
 abstract?
 or does abstract do something completely different?  or is it even
 implemented ;)






-- 
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
Jul 12 2004
prev sibling parent Stewart Gordon <smjg_1998 yahoo.com> writes:
Charlie wrote:


 It only works on methods, not entire classes.


It's been said before.  If one declares a class to be abstract, it 
follows that one wants it to be abstract.  Why is the keyword 'abstract' 
allowed here if it does nothing?


 For that i suggest an interface,
 though it does seem to be either a flaw, or not documented well.


<snip top of upside-down reply>

I'd call it a bug one way or the other.

Stewart.

-- 
My e-mail is valid but not my primary mailbox, aside from its being the 
unfortunate victim of intensive mail-bombing at the moment.  Please keep 
replies on the 'group where everyone may benefit.
Jul 13 2004
prev sibling next sibling parent Regan Heath <regan netwin.co.nz> writes:
On Mon, 12 Jul 2004 13:54:31 -0400, Jarrett Billingsley 
<kb3ctd2 yahoo.com> wrote:


 in C++, there is a concept called an "abstract class", which is achieved 
 by
 having a class with at least one pure virtual function (function 
 declaration
 =0).  you are not allowed to create any instances of an abstract class.

 i thought the keyword "abstract" would cause this behavior on a class
 without the "implied" kind that you get with C++.  apparently not!

 defining a class as "abstract" does not seem to do anything.  D allows 
 you
 to create instances of it.  deriving from the abstract class does not 
 change
 any behavior.

 looking through the manual, i cannot find any explanation as to what the
 abstract keyword does.

 all i'm doing is this:

 abstract class Parent
 {
     int foo() { return 10; }
 }

 class Child : Parent
 {
     int foo() { return 20; }
 }

 i'm still allowed to create instances of Parent and call its foo().

 am i doing something wrong here?  am i not supposed to have any kind of
 function declaration in the Parent class for it to be treated as 
 abstract?
 or does abstract do something completely different?  or is it even
 implemented ;)


It is implemented, it just behaves differently to what you're expecting. 
 From what I can see..

class Parent
{
	int foo();
}

class Child : Parent
{
	int foo() { return 20; }
}

The above will not compile, *unless* I put 'abstract' in front of the 
class, or the method 'foo' eg

abstract class Parent
{
	int foo();
}

or

class Parent
{
	abstract int foo();
}

so abstract *allows* a class to have undefined method bodies. It does not 
*prohibit* instantiation of a class as long as it has all it's method 
bodies.

I *do* like the fact that you do not need to place abstract on all the 
methods of a class, one abstract on the class definition will affect all 
methods in the class.

I *do not* like the fact that you cannot provide bodies and still 
*prohibit* instantiation of the abstract class.

These are my gut reactions, maybe it's time for a rethink, perhaps D has 
it right, if a class has all it's method bodies, why prohibit it's 
instantiation? In which case placing abstract in front of the class or 
method (with a body) should give an error.

Regan.

-- 
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
Jul 12 2004
prev sibling parent "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
thanks for the clarification :)  didn't know if i was just thick or what!

unfortunately an interface isn't really useable here.  so oh well.
Jul 12 2004