www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Bug: help me!

reply "Miguel Ferreira Simões" <Kobold netcabo.pt> writes:
I am not sure if this a dmd bug or if I am bugging...
The following code gives "Access Violation" and I don't unserstand why.
If I remove the "void bugMethod()" in interface ITest the code works.
If I change ITest to an abstract class with an "abstract void bugMethod()" 
the code works too.

Can someone explain me...thanks!

interface ITest {
    void bugMethod();
}

class Test : ITest {
    private IManager _manager;

     public this(IManager manager) {
        _manager = manager;
        _manager.register(this);
    }
    public void bugMethod() {

    }
}

interface IManager {
    void register(ITest itest);
}

class TestManager : IManager {
    private ITest[ITest] _aa;

    public this() {
        new Test(this);
        new Test(this);
    }

    public void register(ITest i) {
        _aa[i] = i;
    }
}

-- 
Miguel Ferreira Simoes 
Nov 28 2004
parent reply Sjoerd van Leent <svanleent wanadoo.nl> writes:
Not quite a bug. It has to do with the list you are using. You're trying 
to have a map with self references, this isn't working.

The following code is working as expected (though the register function 
is different now):

interface ITest {
     void bugMethod();
}

class Test : ITest {
     private IManager _manager;

     public this(IManager manager) {
         _manager = manager;
         _manager.register(this);
     }
     public void bugMethod() {

     }
}

interface IManager {
     uint register(ITest itest);
}

class TestManager : IManager {
     private ITest[ulong] _aa;

     public this() {
         new Test(this);
         //new Test(this);
     }

     public uint register(ITest i)
         static ulong registerInstance;
         _aa[registerInstance] = i;
	++registerInstance;
         return registerInstance - 1;
     }
}

int main(char[][] args) {
	new TestManager();
	return 0;
}

Regards,
Sjoerd

Miguel Ferreira Simões wrote:
 I am not sure if this a dmd bug or if I am bugging...
 The following code gives "Access Violation" and I don't unserstand why.
 If I remove the "void bugMethod()" in interface ITest the code works.
 If I change ITest to an abstract class with an "abstract void bugMethod()" 
 the code works too.
 
 Can someone explain me...thanks!
 
 ...
 

Nov 28 2004
next sibling parent "Miguel Ferreira Simões" <Kobold netcabo.pt> writes:
Thanks!

You said that it's not a bug and that self-referencing maps aren't working. 
But should they work? If so, imho, it seems a bug. Do you know why it works 
with abstract classes (instead of interfaces)? 
Nov 28 2004
prev sibling parent reply Ant <duitoolkit yahoo.ca> writes:
On Mon, 29 Nov 2004 00:13:57 +0100, Sjoerd van Leent wrote:

 Not quite a bug. It has to do with the list you are using. You're trying 
 to have a map with self references, this isn't working.

I don't like that explanation! this works: ############ class NewClass { this() { printf("NewClass"); } } void main() { //new TestManager(); NewClass[NewClass] nn; NewClass a = new NewClass(); nn[a] = a; } ############# Did I not understant what you're saying? and Miguel says it work's if the interface doesn't define one of the methods... I think this is a bug. BTW this type of construct where you use "this" on the constructor is illegal on java (I think...) Ant
Nov 28 2004
parent reply "Simon Buchan" <currently no.where> writes:
On Sun, 28 Nov 2004 19:11:45 -0500, Ant <duitoolkit yahoo.ca> wrote:

 On Mon, 29 Nov 2004 00:13:57 +0100, Sjoerd van Leent wrote:

 Not quite a bug. It has to do with the list you are using. You're trying
 to have a map with self references, this isn't working.

I don't like that explanation! this works: ############ class NewClass { this() { printf("NewClass"); } } void main() { //new TestManager(); NewClass[NewClass] nn; NewClass a = new NewClass(); nn[a] = a; } ############# Did I not understant what you're saying? and Miguel says it work's if the interface doesn't define one of the methods... I think this is a bug. BTW this type of construct where you use "this" on the constructor is illegal on java (I think...) Ant

What about D :D T[T] should be fine... after all, we have all done dictionaries, right? (char[][char[]] dictionary;) Other than that... well, lets just say OO isn't ANYONE'S strong point. -- Using Opera's revolutionary e-mail client: http://www.opera.com/m2/
Nov 28 2004
parent reply Ant <duitoolkit yahoo.ca> writes:
On Mon, 29 Nov 2004 18:38:38 +1300, Simon Buchan wrote:

 On Sun, 28 Nov 2004 19:11:45 -0500, Ant <duitoolkit yahoo.ca> wrote:
 
 
 What about D :D
 
 T[T] should be fine... after all, we have all done dictionaries, right?
 (char[][char[]] dictionary;)
 
 Other than that... well, lets just say OO isn't ANYONE'S strong point.

??? I don't get it... what are you saying? I'm thinking I didn't understant Sjoerd van Leent explanation... Ant
Nov 28 2004
parent reply "Simon Buchan" <currently no.where> writes:
On Mon, 29 Nov 2004 00:53:07 -0500, Ant <duitoolkit yahoo.ca> wrote:

 On Mon, 29 Nov 2004 18:38:38 +1300, Simon Buchan wrote:

 On Sun, 28 Nov 2004 19:11:45 -0500, Ant <duitoolkit yahoo.ca> wrote:


 What about D :D

 T[T] should be fine... after all, we have all done dictionaries, right?
 (char[][char[]] dictionary;)

 Other than that... well, lets just say OO isn't ANYONE'S strong point.

??? I don't get it... what are you saying? I'm thinking I didn't understant Sjoerd van Leent explanation... Ant

Sorry... What I meant was: What about using this in a D constructor, (as opposed to Java) is that OK? You are right that it is unlikely a self-typed assosiative array would be the cause (even if the value was the key) (BTW, is void[<sometype>] allowed as a set type? Should it be?) Object Oriented programming is confusing to everybody, even the ppl who write compilers and invent languages, so what chance do WE have? Is that clearer? I'm thinking this case has to be reduced first... -- Using Opera's revolutionary e-mail client: http://www.opera.com/m2/
Nov 28 2004
parent reply Ant <duitoolkit yahoo.ca> writes:
On Mon, 29 Nov 2004 20:24:27 +1300, Simon Buchan wrote:

 On Mon, 29 Nov 2004 00:53:07 -0500, Ant <duitoolkit yahoo.ca> wrote:
 
 On Mon, 29 Nov 2004 18:38:38 +1300, Simon Buchan wrote:

 On Sun, 28 Nov 2004 19:11:45 -0500, Ant <duitoolkit yahoo.ca> wrote:


 What about D :D

 T[T] should be fine... after all, we have all done dictionaries, right?
 (char[][char[]] dictionary;)

 Other than that... well, lets just say OO isn't ANYONE'S strong point.

??? I don't get it... what are you saying? I'm thinking I didn't understant Sjoerd van Leent explanation... Ant

Sorry... What I meant was: What about using this in a D constructor, (as opposed to Java) is that OK?

it's suppose to be. java, as usual - I guess, doesn't allow it for simplicity on the implementation. D, as usual, is smarter, and, as usual, buggier (it's beta after all). what strikes me is the interface. I didn't try it but seems it works fine without the method defined on the interface. If changing the interface makes the executable valid or not then it's a bug anyway we look at it.
 
 Object Oriented programming is confusing to everybody, even the ppl
 who write compilers and invent languages, so what chance do WE have?
 
 Is that clearer?

thanks :)
 
 I'm thinking this case has to be reduced first...

Ant
Nov 28 2004
parent Sjoerd van Leent <svanleent wanadoo.nl> writes:
Ant wrote:
 On Mon, 29 Nov 2004 20:24:27 +1300, Simon Buchan wrote:
 
 
On Mon, 29 Nov 2004 00:53:07 -0500, Ant <duitoolkit yahoo.ca> wrote:


On Mon, 29 Nov 2004 18:38:38 +1300, Simon Buchan wrote:


On Sun, 28 Nov 2004 19:11:45 -0500, Ant <duitoolkit yahoo.ca> wrote:


What about D :D

T[T] should be fine... after all, we have all done dictionaries, right?
(char[][char[]] dictionary;)

Other than that... well, lets just say OO isn't ANYONE'S strong point.

??? I don't get it... what are you saying? I'm thinking I didn't understant Sjoerd van Leent explanation... Ant

Sorry... What I meant was: What about using this in a D constructor, (as opposed to Java) is that OK?

it's suppose to be. java, as usual - I guess, doesn't allow it for simplicity on the implementation. D, as usual, is smarter, and, as usual, buggier (it's beta after all). what strikes me is the interface. I didn't try it but seems it works fine without the method defined on the interface. If changing the interface makes the executable valid or not then it's a bug anyway we look at it.
Object Oriented programming is confusing to everybody, even the ppl
who write compilers and invent languages, so what chance do WE have?

Is that clearer?

thanks :)
I'm thinking this case has to be reduced first...

Ant

Allright, it is a bug (or not, it may be intented behaviour, but I don't think it is). And I didn't mean to say that self references of interfaces with a method shouldn't work, I just doesn't seem to work. As I read over the part of stating that interfaces without a method do work as expected. This is more weird since their are more fields then only the methods specified. My apologies for being unclear. Regards, Sjoerd
Nov 29 2004