• "Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> Dec 07 2009
• KennyTM~ <kennytm gmail.com> Dec 07 2009
• Ali Cehreli <acehreli yahoo.com> Dec 07 2009
• Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> Dec 07 2009
• "Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> Dec 08 2009

"Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> writes:

The decision to dump the built-in complex and imaginary types in favour 
of std.complex.Complex was made a long time ago, but nothing has 
happened yet. So I have a few questions:

1. Is this still planned for D2?

2. In that case, when will it happen?

3. Are there any reasons why I shouldn't replace cxxx with Complex!xxx 
in my code right away?

-Lars

KennyTM~ <kennytm gmail.com> writes:

On Dec 7, 09 22:40, Lars T. Kyllingstad wrote:
 The decision to dump the built-in complex and imaginary types in favour
 of std.complex.Complex was made a long time ago, but nothing has
 happened yet. So I have a few questions:

 1. Is this still planned for D2?

 2. In that case, when will it happen?

 3. Are there any reasons why I shouldn't replace cxxx with Complex!xxx
 in my code right away?


All opBinary-s (opAdd, etc.) aren't implemented now, so e.g. z+w won't 
compile for Complex!T.

 -Lars

Ali Cehreli <acehreli yahoo.com> writes:

Lars T. Kyllingstad Wrote:

 The decision to dump the built-in complex and imaginary types in favour 
 of std.complex.Complex was made a long time ago, but nothing has 
 happened yet. So I have a few questions:
 
 1. Is this still planned for D2?
 
 2. In that case, when will it happen?
 
 3. Are there any reasons why I shouldn't replace cxxx with Complex!xxx 
 in my code right away?
 
 -Lars


An introductory article on Digital Mars explains the rationale behind the
complex types and quotes W. Kahan: "The language of pairs is incorrect for
Complex Arithmetic; it needs the Imaginary type."

  http://www.digitalmars.com/d/2.0/cppcomplex.html

The library types can and do work as well, probably because "Appendix G of the
C99 standard has recommendations for dealing with this problem."

Ali

Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:

Lars T. Kyllingstad wrote:
 The decision to dump the built-in complex and imaginary types in favour 
 of std.complex.Complex was made a long time ago, but nothing has 
 happened yet. So I have a few questions:
 
 1. Is this still planned for D2?


Far as I know, yes.

 2. In that case, when will it happen?


I don't know.

 3. Are there any reasons why I shouldn't replace cxxx with Complex!xxx 
 in my code right away?


I think we'll have to provide aliases for the existing complex types, so 
your code should continue functioning after the change.


Andrei

"Lars T. Kyllingstad" <public kyllingen.NOSPAMnet> writes:

Andrei Alexandrescu wrote:
 Lars T. Kyllingstad wrote:
 The decision to dump the built-in complex and imaginary types in 
 favour of std.complex.Complex was made a long time ago, but nothing 
 has happened yet. So I have a few questions:

 1. Is this still planned for D2?


 Far as I know, yes.
 
 2. In that case, when will it happen?


 I don't know.
 
 3. Are there any reasons why I shouldn't replace cxxx with Complex!xxx 
 in my code right away?


 I think we'll have to provide aliases for the existing complex types, so 
 your code should continue functioning after the change.


The big difference is in assignments:

   real a, b;
   creal x = a + b*1.0i;
   auto y = Complex!real(a, b);

   // Also, this doesn't work with current complex types, because
   // creal.re and creal.im aren't lvalues:
   Complex!real z;
   z.re = a;
   z.im = b;

But it's not a problem, really. Since Complex!T is binary compatible 
with the current complex types I can just insert a few casts here and 
there. I just wanted to make sure I wouldn't have to go back to creal & 
co later.

-Lars