• deed (19/19) Jun 19 2013 The following compiles and crashes with DMD 2.063.
• bearophile (5/6) Jun 19 2013 In general it's not easy for a D compiler to perform that kind of
• Iain Buclaw (4/13) Jun 19 2013 Isn't the problem in this property function? (Shouldn't it
• deed (1/8) Jun 19 2013 that's also an error. changing to _var gives same result though..
• dennis luehring (5/21) Jun 19 2013 that isn't the problem - D allows assignment to an read property - and
• dennis luehring (1/5) Jun 19 2013 sorry i've totaly lost seeing the write property :(
• John Colvin (8/9) Jun 19 2013 Yes it should, in an ideal world. Expanding out the assign

"deed" <none none.none> writes:

The following compiles and crashes with DMD 2.063.
Should this be a compile time error?


class A
{
     int _var;

     void var(int i)  property
     {
         this.var = i;   // oops, crashes.
     }                   // should have been this._var

     int var()  property
     {
         return var;
     }
}

void main()
{
     auto a = new A();
     a.var = 3;
}

"bearophile" <bearophileHUGS lycos.com> writes:

deed:

 Should this be a compile time error?

In general it's not easy for a D compiler to perform that kind of 
validation.

Bye,
bearophile

"Iain Buclaw" <ibuclaw ubuntu.com> writes:

On Wednesday, 19 June 2013 at 11:33:43 UTC, deed wrote:
 The following compiles and crashes with DMD 2.063.
 Should this be a compile time error?


 class A
 {
     int _var;

/* SNIP */

     int var()  property
     {
         return var;
     }

Isn't the problem in this property function?  (Shouldn't it 
return _var :o)

"deed" <none none.none> writes:

 /* SNIP */

    int var()  property
    {
        return var;
    }

 Isn't the problem in this property function?  (Shouldn't it 
 return _var :o)

that's also an error. changing to _var gives same result though..

dennis luehring <dl.soluz gmx.net> writes:

Am 19.06.2013 13:40, schrieb Iain Buclaw:
 On Wednesday, 19 June 2013 at 11:33:43 UTC, deed wrote:
 The following compiles and crashes with DMD 2.063.
 Should this be a compile time error?


 class A
 {
     int _var;

 /* SNIP */

     int var()  property
     {
         return var;
     }

 Isn't the problem in this property function?  (Shouldn't it
 return _var :o)

that isn't the problem - D allows assignment to an read property - and 
there is no write property around, so it should be an compiletime error

i should compile only if the missing write property is available - or?

 property int var(int value) { return _var = value; }

dennis luehring <dl.soluz gmx.net> writes:

 that isn't the problem - D allows assignment to an read property - and
 there is no write property around, so it should be an compiletime error

 i should compile only if the missing write property is available - or?

  property int var(int value) { return _var = value; }

sorry i've totaly lost seeing the write property :(

"John Colvin" <john.loughran.colvin gmail.com> writes:

On Wednesday, 19 June 2013 at 11:33:43 UTC, deed wrote:
 Should this be a compile time error?

Yes it should, in an ideal world. Expanding out the assign 
property, what we've got is:

void var(int i)
{
      var(i);   // endless recursion.
}

Linux segfaults on stack overflow, so that's your crash.