• #ponce <spam spam.spam> Apr 23 2010
• Robert Clipsham <robert octarineparrot.com> Apr 23 2010
• Philippe Sigaud <philippe.sigaud gmail.com> Apr 23 2010

#ponce <spam spam.spam> writes:

In C++ implicit constructors and conversion operators allow a user-defined type
to act quite like a builtin-type.

  struct half
  {
      half(float x);l
      inline operator float() const;
  }

allows to write:

  half x = 1.f;
  float f = x;

and this is especially useful for templates. 
I couldn't find a similar construct in D, is there any?

Robert Clipsham <robert octarineparrot.com> writes:

On 23/04/10 17:22, #ponce wrote:
 In C++ implicit constructors and conversion operators allow a user-defined
type to act quite like a builtin-type.

    struct half
    {
        half(float x);l
        inline operator float() const;
    }

 allows to write:

    half x = 1.f;
    float f = x;

 and this is especially useful for templates.
 I couldn't find a similar construct in D, is there any?


A combination of alias this and opAssign should work:
----
struct half
{
   float f;
   alias f this;
   half opAssign(float fl)
   {
     f = fl;
     return this;
   }
}

void main()
{
   half x;
   // For some reason I got a compilation error with half x = 1f;
   x = 1f;
   float f = x;
}
----

Philippe Sigaud <philippe.sigaud gmail.com> writes:

--00032555ad522b39650484edb99a
Content-Type: text/plain; charset=ISO-8859-1

On Fri, Apr 23, 2010 at 18:46, Robert Clipsham <robert octarineparrot.com>wrote:

 On 23/04/10 17:22, #ponce wrote:

 In C++ implicit constructors and conversion operators allow a user-defined
 type to act quite like a builtin-type.

   struct half
   {
       half(float x);l
       inline operator float() const;
   }

 allows to write:

   half x = 1.f;
   float f = x;

 and this is especially useful for templates.
 I couldn't find a similar construct in D, is there any?


 A combination of alias this and opAssign should work:
 ----
 struct half
 {
  float f;
  alias f this;
  half opAssign(float fl)
  {
    f = fl;
    return this;
  }
 }

 void main()
 {
  half x;
  // For some reason I got a compilation error with half x = 1f;
  x = 1f;
  float f = x;
 }


Could the compilation error comes from the compiler rewriting half f = 1f;
into half f(1f); or somesuch and trying to call a constructor?
I'm tearing my hair due to errors/confusions between constructors and opCall
for structs.

Anyway, in your example, adding

this(float f) { this.f = f;}

is enough to get half f = 1f; to compile.

--00032555ad522b39650484edb99a
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div class=3D"gmail_quote">On Fri, Apr 23, 2010 at 18:46, Robert Clipsham <=
span dir=3D"ltr">&lt;<a href=3D"mailto:robert octarineparrot.com">robert oc=
tarineparrot.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote"=
 style=3D"margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 2=
04); padding-left: 1ex;">
<div><div></div><div class=3D"h5">On 23/04/10 17:22, #ponce wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin: 0pt 0pt 0pt 0.8ex; borde=
r-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
In C++ implicit constructors and conversion operators allow a user-defined =
type to act quite like a builtin-type.<br>
<br>
 =A0 struct half<br>
 =A0 {<br>
 =A0 =A0 =A0 half(float x);l<br>
 =A0 =A0 =A0 inline operator float() const;<br>
 =A0 }<br>
<br>
allows to write:<br>
<br>
 =A0 half x =3D 1.f;<br>
 =A0 float f =3D x;<br>
<br>
and this is especially useful for templates.<br>
I couldn&#39;t find a similar construct in D, is there any?<br>
</blockquote>
<br></div></div>
A combination of alias this and opAssign should work:<br>
----<br>
struct half<br>
{<br>
 =A0float f;<br>
 =A0alias f this;<br>
 =A0half opAssign(float fl)<br>
 =A0{<br>
 =A0 =A0f =3D fl;<br>
 =A0 =A0return this;<br>
 =A0}<br>
}<br>
<br>
void main()<br>
{<br>
 =A0half x;<br>
 =A0// For some reason I got a compilation error with half x =3D 1f;<br>
 =A0x =3D 1f;<br>
 =A0float f =3D x;<br>
}<br></blockquote><div><br>Could the compilation error comes from the compi=
ler rewriting half f =3D 1f; into half f(1f); or somesuch and trying to cal=
l a constructor?<br>I&#39;m tearing my hair due to errors/confusions betwee=
n constructors and opCall for structs.<br>
<br>Anyway, in your example, adding <br><br>this(float f) { this.f =3D f;}<=
br><br>is enough to get half f =3D 1f; to compile.<br><br></div></div>

--00032555ad522b39650484edb99a--