i'm liking D, but i came across this .... ugliness.

class A
{
 int x;
}

class B
{
 int y;
}

int main()
{
 A a=new A;
 B b=cast(B) a;
 return 0;
}

trying to access members of b will cause a memory access violation.

why am i allowed to do this?

Jarrett Billingsley wrote:
> i'm liking D, but i came across this .... ugliness.
> 
> class A
> {
>  int x;
> }
> 
> class B
> {
>  int y;
> }
> 
> int main()
> {
>  A a=new A;
>  B b=cast(B) a;
>  return 0;
> }
> 
> trying to access members of b will cause a memory access violation.
> 
> why am i allowed to do this?
> 
> 
When a cast fails, the result is null. So in this case you can catch the error with a simple test:

if(b === null)
	printf("B is null");

or alternatively use the 'is' operator, which is the same as '===':

if(b is null)
 ...

Mike Parker wrote:
> When a cast fails, the result is null. So in this case you can catch the error with a simple test:
> 
> if(b === null)
>     printf("B is null");
> 
> or alternatively use the 'is' operator, which is the same as '===':
> 
> if(b is null)
>  ...

Right, but it's pretty strange that it's not a compile-time error to cast an A to a B, even though B does not inherit A.  The cast *has* to fail, so why is it even allowed?

 -- andy

Andy Friesen wrote:

> Mike Parker wrote:
>> When a cast fails, the result is null. So in this case you can catch the error with a simple test:
>> 
>> if(b === null)
>>     printf("B is null");
>> 
>> or alternatively use the 'is' operator, which is the same as '===':
>> 
>> if(b is null)
>>  ...
> 
> Right, but it's pretty strange that it's not a compile-time error to cast an A to a B, even though B does not inherit A.  The cast *has* to fail, so why is it even allowed?

True. The specifications are not very specific in this point (Heyheyhey,
what a pun!!!):

"D differs from C/C++ in another aspect of casts. Any casting of a class reference to a derived class reference is done with a runtime check to make sure it really is a proper downcast. This means that it is equivalent to the behavior of the dynamic_cast operator in C++."

A sentence should be added:

"Casts between references to unrelated classes references are illegal."

In article <ccalsv$c9v$1@digitaldaemon.com>, Mike Parker says...

>When a cast fails, the result is null. So in this case you can catch the error with a simple test:

But why should we have to add run-time code for something that could be detected at compile-time? This is a compile-time error. It should be reported at compile-time.

Arcane Jill

"Arcane Jill" <Arcane_member@pathlink.com> wrote in message news:ccb2t5$10tm$1@digitaldaemon.com...
> In article <ccalsv$c9v$1@digitaldaemon.com>, Mike Parker says...
>
> >When a cast fails, the result is null. So in this case you can catch the error with a simple test:
>
> But why should we have to add run-time code for something that could be
detected
> at compile-time? This is a compile-time error. It should be reported at compile-time.

It's a runtime error because one may be asking this question as part of the logic of the implementation of a template that selects one of several alternatives based on the type of a parameter.

Walter wrote:
> "Arcane Jill" <Arcane_member@pathlink.com> wrote in message
> news:ccb2t5$10tm$1@digitaldaemon.com...
> 
>>In article <ccalsv$c9v$1@digitaldaemon.com>, Mike Parker says...
>>
>>
>>>When a cast fails, the result is null. So in this case you can catch the
>>>error with a simple test:
>>
>>But why should we have to add run-time code for something that could be
> 
> detected
> 
>>at compile-time? This is a compile-time error. It should be reported at
>>compile-time.
> 
> 
> It's a runtime error because one may be asking this question as part of the
> logic of the implementation of a template that selects one of several
> alternatives based on the type of a parameter.
> 
> 
You're supposed to do that through template specialization, not run time checks :-), no?

"Daniel Horn" <hellcatv@hotmail.com> wrote in message news:ccc6jd$2t30$1@digitaldaemon.com...
> Walter wrote:
> > It's a runtime error because one may be asking this question as part of
the
> > logic of the implementation of a template that selects one of several alternatives based on the type of a parameter.
> You're supposed to do that through template specialization, not run time checks :-), no?

Trying to do everything with specialization has been listed as one of the problems with C++ generic programming by some people who have made their reputations writing template libraries, so I am inclined to believe them <g>.