More Double/Float merging

[Mark Wielaard <mark at klomp dot org>]

Hi,

Here are the suggestions from the Classpath list for the remaining
Double/Float merge. This shows no new regressions with the Double/Float
tests in Mauve.

With this patch Float is completely merged with Classpath, Double still
has a different (native) parseDouble() method. This needs working on the
Classpath side.

2001-10-19  Mark Wielaard  <mark@klomp.org>

	* java/lang/Double.java: More Classpath merging
	(isInfinite): Don't use doubleToLongBits
	(isNaN (Object)): return v != v
	(initIDs): make native
	* java/lang/Float.java: Ditto
	(isInfinite): Don't use floatToIntBits
	(isNaN (Object)): return v != v
	* java/lang/natDouble.cc: add empty initIDs()

OK to commit?

Cheers,

Mark
-- 
Stuff to read:
    <http://www.toad.com/gnu/whatswrong.html>
  What's Wrong with Copy Protection, by John Gilmore
Index: java/lang/Double.java
===================================================================
RCS file: /cvs/gcc/gcc/libjava/java/lang/Double.java,v
retrieving revision 1.10
diff -u -r1.10 Double.java
--- Double.java	2001/10/15 06:31:13	1.10
+++ Double.java	2001/10/20 01:09:27
@@ -146,6 +146,14 @@
    * <code>instanceof</code> <code>Double</code>, and represents
    * the same primitive <code>double</code> value return 
    * <code>true</code>.  Otherwise <code>false</code> is returned.
+   * <p>
+   * Note that there are two differences between <code>==</code> and
+   * <code>equals()</code>. <code>0.0d == -0.0d</code> returns <code>true</code>
+   * but <code>new Double(0.0d).equals(new Double(-0.0d))</code> returns
+   * <code>false</code>. And <code>Double.NaN == Double.NaN</code> returns
+   * <code>false</code>, but
+   * <code>new Double(Double.NaN).equals(new Double(Double.NaN))</code> returns
+   * <code>true</code>.
    *
    * @param obj the object to compare to
    * @return whether the objects are semantically equal.
@@ -248,11 +256,9 @@
    */
   public static boolean isNaN (double v)
   {
-    long bits = doubleToLongBits (v);
-    long e = bits & 0x7ff0000000000000L;
-    long f = bits & 0x000fffffffffffffL;
-
-    return e == 0x7ff0000000000000L && f != 0L;
+    // This works since NaN != NaN is the only reflexive inequality
+    // comparison which returns true.
+    return v != v;
   }
 
   /**
@@ -277,10 +283,7 @@
    */
   public static boolean isInfinite (double v)
   {
-    long bits = doubleToLongBits (v);
-    long f = bits & 0x7fffffffffffffffL;
-
-    return f == 0x7ff0000000000000L;
+    return (v == POSITIVE_INFINITY || v == NEGATIVE_INFINITY);
   }
 
   /**
@@ -508,5 +511,5 @@
    * Initialize JNI cache.  This method is called only by the 
    * static initializer when using JNI.
    */
-  private static void initIDs () { /* Not used in libgcj */ };
+  private static native void initIDs ();
 }
Index: java/lang/Float.java
===================================================================
RCS file: /cvs/gcc/gcc/libjava/java/lang/Float.java,v
retrieving revision 1.9
diff -u -r1.9 Float.java
--- Float.java	2001/10/15 06:31:13	1.9
+++ Float.java	2001/10/20 01:09:27
@@ -215,6 +215,14 @@
    * <code>instanceof</code> <code>Float</code>, and represents
    * the same primitive <code>float</code> value return 
    * <code>true</code>.  Otherwise <code>false</code> is returned.
+   * <p>
+   * Note that there are two differences between <code>==</code> and
+   * <code>equals()</code>. <code>0.0f == -0.0f</code> returns <code>true</code>
+   * but <code>new Float(0.0f).equals(new Float(-0.0f))</code> returns
+   * <code>false</code>. And <code>Float.NaN == Float.NaN</code> returns
+   * <code>false</code>, but
+   * <code>new Float(Float.NaN).equals(new Float(Float.NaN))</code> returns
+   * <code>true</code>.
    *
    * @param obj the object to compare to
    * @return whether the objects are semantically equal.
@@ -364,11 +372,9 @@
    */
   public static boolean isNaN (float v)
   {
-    int bits = floatToIntBits (v);
-    int e = bits & 0x7f800000;
-    int f = bits & 0x007fffff;
-
-    return e == 0x7f800000 && f != 0;
+    // This works since NaN != NaN is the only reflexive inequality
+    // comparison which returns true.
+    return v != v;
   }
 
   /**
@@ -393,10 +399,7 @@
    */
   public static boolean isInfinite (float v)
   {
-    int bits = floatToIntBits (v);
-    int f = bits & 0x7fffffff;
-
-    return f == 0x7f800000;
+    return (v == POSITIVE_INFINITY || v == NEGATIVE_INFINITY);
   }
 
   /**
Index: java/lang/natDouble.cc
===================================================================
RCS file: /cvs/gcc/gcc/libjava/java/lang/natDouble.cc,v
retrieving revision 1.14
diff -u -r1.14 natDouble.cc
--- natDouble.cc	2001/03/26 07:05:32	1.14
+++ natDouble.cc	2001/10/20 01:09:27
@@ -186,3 +186,9 @@
     }
   throw new NumberFormatException;
 }
+
+void
+java::lang::Double::initIDs()
+{
+  // Not used in libgcj
+}