Hello,

I would do something like this, but this return me an execution error :
object.Exception: Stack Overflow

// --------CODE--------
class A
{
 void delegate() dg;

 void doIt()
 {
  dg();
 }
}

class B
{
 A a;

 this()
 {
  a = new A;
  a.dg = { doSomething(); };
 }

 void doSomething() { }
}

void main()
{
 auto b = new B;
 b.a.doIt();
}
// ------END CODE------

Is this a bug or have I do something wrong ?
Thanks in advance for your help,
TSalm

Reply to TSalm,

> Hello,
> 
> I would do something like this, but this return me an execution error
> : object.Exception: Stack Overflow
> 
> // --------CODE--------
> class A
> {
> void delegate() dg;
> void doIt()
> {
> dg();
> }
> }
> class B
> {
> A a;
> this()
> {
> a = new A;
> a.dg = { doSomething(); };
> }
> void doSomething() { }
> }
> void main()
> {
> auto b = new B;
> b.a.doIt();
> }
> // ------END CODE------
> 
> Is this a bug or have I do something wrong ?
> Thanks in advance for your help,
> TSalm

If this is d1.0 the error is that you are allowing an anon delegate to escape the enclosing function.

a.dg = &this.doSomething;  // this would be ok if that helps.


if it is 2.0, I think this is correct.

Le Sat, 29 Nov 2008 01:08:28 +0100, BCS <ao@pathlink.com> a écrit:

> Reply to TSalm,
>
>> Hello,
>>  I would do something like this, but this return me an execution error
>> : object.Exception: Stack Overflow
>>  // --------CODE--------
>> class A
>> {
>> void delegate() dg;
>> void doIt()
>> {
>> dg();
>> }
>> }
>> class B
>> {
>> A a;
>> this()
>> {
>> a = new A;
>> a.dg = { doSomething(); };
>> }
>> void doSomething() { }
>> }
>> void main()
>> {
>> auto b = new B;
>> b.a.doIt();
>> }
>> // ------END CODE------
>>  Is this a bug or have I do something wrong ?
>> Thanks in advance for your help,
>> TSalm
>
> If this is d1.0 the error is that you are allowing an anon delegate to escape the enclosing function.

Yes, it's on D1.036.


>
> a.dg = &this.doSomething;  // this would be ok if that helps.
>
>
>  if it is 2.0, I think this is correct.
>

Yes, you are right.
But this is an example code. The "true" code uses delegates with argument which differs from called functions, so I can't point delegate directly to them.
And this anonymous function's way is really fastest to code... I must waiting for a D2 stable version ;-)

Thanks

Reply to TSalm,

> Yes, you are right.
> But this is an example code. The "true" code uses delegates with
> argument
> which differs from called functions, so I can't point delegate
> directly to
> them.
> And this anonymous function's way is really fastest to code... I must
> waiting for a D2 stable version ;-)
> Thanks
> 

you can store off the values you need like this:


struct C(R, A...)
{
  A args;
  R function(A) dg;
  static R delegate() opCall(R function(A) dg, A a)
  {
     C!(R, A) ret;
     ret.dg=dg;
     foreach(int i,_;A)
        ret.args[i] = a[i];
     return &ret.fn;
  }
  R fn() { return dg(args); }
}


// test it

import std.stdio;

int delegate() dg(int i)
{
  return C!(int,int)(function int(int j){return j;}, i);   // <--- used here
}

void main()
{
  auto d = dg(5);
  writef("%s\n", d());
}

Le Sat, 29 Nov 2008 01:49:20 +0100, BCS <ao@pathlink.com> a écrit:

> struct C(R, A...)
> {
>   A args;
>   R function(A) dg;
>   static R delegate() opCall(R function(A) dg, A a)
>   {
>      C!(R, A) ret;
>      ret.dg=dg;
>      foreach(int i,_;A)
>         ret.args[i] = a[i];
>      return &ret.fn;
>   }
>   R fn() { return dg(args); }
> }
>   // test it
>  import std.stdio;
>  int delegate() dg(int i)
> {
>   return C!(int,int)(function int(int j){return j;}, i);   // <--- used here
> }
>  void main()
> {
>   auto d = dg(5);
>   writef("%s\n", d());
> }

Interesting code. Thanks!