[PATCH] Complex.polar(x,0) should produce a RealNum

[PATCH] Complex.polar(x,0) should produce a RealNum

[Jamison Hope]

[-- Attachment #1: Type: text/plain, Size: 627 bytes --]

This patch changes Complex#polar so that if its second argument (the
angle) is exact zero, then the result is a RealNum.  This makes it
consistent with how Complex#make works.

Otherwise you end up with the strange situation of:

#|kawa:1|# (make-polar 3 0)
3.0
#|kawa:2|# (= (make-polar 3 0) 3.0)
#t
#|kawa:3|# (real? 3.0)
#t
#|kawa:4|# (real? (make-polar 3 0))
#f
#|kawa:5|# (make-polar 3 0):class
class gnu.math.DComplex


N.B. (real? (make-polar 3 0)) returns #t in all the other Schemes I
checked (Gambit, Guile, MIT, and Racket), but doing so is not
required by RnRS.

--
Jamison Hope
The PTR Group
www.theptrgroup.com



[-- Attachment #2: polarzero.patch --]
[-- Type: application/octet-stream, Size: 1692 bytes --]

Index: gnu/math/ChangeLog
===================================================================
--- gnu/math/ChangeLog	(revision 8230)
+++ gnu/math/ChangeLog	(working copy)
@@ -1,3 +1,9 @@
+2014-12-18  Jamison Hope  <jrh@theptrgroup.com>
+
+	* Complex.java (polar): Change return type from DComplex to
+	Complex and return RealNum if angle is exact zero.
+	* DComplex.java (power): Update return type accordingly.
+
 2014-11-17  Jamison Hope  <jrh@theptrgroup.com>
 
 	Implemented quaternions.
Index: gnu/math/Complex.java
===================================================================
--- gnu/math/Complex.java	(revision 8230)
+++ gnu/math/Complex.java	(working copy)
@@ -110,13 +110,17 @@
     return new DComplex(re, im);
   }
 
-  public static DComplex polar (double r, double t)
+  public static Complex polar (double r, double t)
   {
+    if (t == 0.0)
+      return new DFloNum(r);
     return new DComplex(r * Math.cos(t), r * Math.sin(t));
   }
 
-  public static DComplex polar (RealNum r, RealNum t)
+  public static Complex polar (RealNum r, RealNum t)
   {
+    if (t.isZero() && t.isExact())
+      return r;
     return polar(r.doubleValue(), t.doubleValue());
   }
 
Index: gnu/math/DComplex.java
===================================================================
--- gnu/math/DComplex.java	(revision 8230)
+++ gnu/math/DComplex.java	(working copy)
@@ -126,8 +126,8 @@
     return ((Numeric)y).divReversed(this);
   }
 
-  public static DComplex power (double x_re, double x_im,
-				double y_re, double y_im)
+  public static Complex power (double x_re, double x_im,
+                               double y_re, double y_im)
   {
     double h;
     /* #ifdef JAVA5 */

Re: [PATCH] Complex.polar(x,0) should produce a RealNum

[Per Bothner]

On 12/17/2014 11:00 PM, Jamison Hope wrote:
> This patch changes Complex#polar so that if its second argument (the
> angle) is exact zero, then the result is a RealNum.  This makes it
> consistent with how Complex#make works.

Thanks.  I checked this in.
-- 
	--Per Bothner
per@bothner.com   http://per.bothner.com/