I checked something out today: Can contracts be defined in an interface.  I was a little surprised at what I found out. Contracts in interfaces will compile without a mutter (though you need a semicolon after the in{}out{} blocks in interfaces), but will not actually be compiled into the code.

Test code with contracts in interfaces:

// START
interface test
{
	uint test(uint number)
	in
	{
		assert(number >= 0);
		assert(number <= 1000);
	}
	out (result)
	{
		assert(result >= 0);
		assert(result >= 10);
	};
}

class TestImpl : test
{
	uint test(uint number)
	body
	{
		return(number / 100);
	}
}

int main()
{
	TestImpl foobar = new TestImpl;
	printf("%d -> %d\n", 41, foobar.test(41));
	printf("%d -> %d\n", 441, foobar.test(441));
	printf("%d -> %d\n", 1241, foobar.test(1241));
	printf("%d -> %d\n", -41, foobar.test(-41));
	printf("%d -> %d\n", 500, foobar.test(500));

	return 0;
}
// END

This returns the following:

>dbcinterfaces
41 -> 0
441 -> 4
1241 -> 12
-41 -> 42949672
500 -> 5

Now, by copy pasting the contract into the implemtation, you get the expected result:

>dbcinterfaces
Error: AssertError Failure dbcinterfaces.d(27)

To me it would seem intuitive if it were possible to define contracts within interfaces, so I'm wondering if there's any reason why this is the way it is?

Cheers,
Sigbjørn Lund Olsen

Sigbjørn Lund Olsen wrote:

> -41 -> 42949672

And may I motion that some solution has to be found to the implicit conversion 'problem'? (atm I'm thinking I'll just sprinkle every function in contract with a static assert with typeof to do the job for me, but that's still another thing I have to remember rather than have the language do it, or be capable of doing it, for me)

Cheers,
Sigbjørn Lund Olsen

I'm not sure what the right answer is here.

"Sigbjørn Lund Olsen" <sigbjorn@lundolsen.net> wrote in message news:c3josp$r3a$1@digitaldaemon.com...
> I checked something out today: Can contracts be defined in an interface.
>   I was a little surprised at what I found out. Contracts in interfaces
> will compile without a mutter (though you need a semicolon after the
> in{}out{} blocks in interfaces), but will not actually be compiled into
> the code.
>
> Test code with contracts in interfaces:
>
> // START
> interface test
> {
> uint test(uint number)
> in
> {
> assert(number >= 0);
> assert(number <= 1000);
> }
> out (result)
> {
> assert(result >= 0);
> assert(result >= 10);
> };
> }
>
> class TestImpl : test
> {
> uint test(uint number)
> body
> {
> return(number / 100);
> }
> }
>
> int main()
> {
> TestImpl foobar = new TestImpl;
> printf("%d -> %d\n", 41, foobar.test(41));
> printf("%d -> %d\n", 441, foobar.test(441));
> printf("%d -> %d\n", 1241, foobar.test(1241));
> printf("%d -> %d\n", -41, foobar.test(-41));
> printf("%d -> %d\n", 500, foobar.test(500));
>
> return 0;
> }
> // END
>
> This returns the following:
>
>  >dbcinterfaces
> 41 -> 0
> 441 -> 4
> 1241 -> 12
> -41 -> 42949672
> 500 -> 5
>
> Now, by copy pasting the contract into the implemtation, you get the expected result:
>
>  >dbcinterfaces
> Error: AssertError Failure dbcinterfaces.d(27)
>
> To me it would seem intuitive if it were possible to define contracts within interfaces, so I'm wondering if there's any reason why this is the way it is?
>
> Cheers,
> Sigbjørn Lund Olsen

On Tue, 23 Mar 2004 16:59:55 -0800, Walter wrote:

> I'm not sure what the right answer is here.

I vote that interfaces should be able to have contracts which would be checked on all implementing methods. I think it would be tremendously useful to be able to explicitly specify the contract like this.

An example: many standard Java APIs use interfaces (collections, for example). The contracts on these methods is spelled out explicitly in the documentation but there is no enforcement.

An interface usually has a contract to go with it, even if it's not made explicit.

Mike Swieton
__
Anyone who attempts to generate random numbers by deterministic means is, of
course, living in a state of sin.
	- John Von Neumann