Patch: FYI: Math/StrictMath merge

Boehm, Hans hans_boehm@hp.com
Wed Oct 8 23:09:00 GMT 2003


Could someone explain why this is a good idea?  It seems to replace one
multiplication by two, and it introduces additional overflow possibilities.

I'd be willing to accept both if it improves accuracy.  What's the argument
that it does?

Hans

> from  Tom Tromey  <tromey@redhat.com>
> 
> 	* java/lang/StrictMath.java (toDegrees): Multiply before
> 	dividing.
> 	(toRadians): Likewise.
> 
> 2003-10-08  C. Brian Jones  <cbj@gnu.org>
> 
> 	* java/lang/Math.java
> 	(toRadians): multiply before dividing to reduce decimal error
> 	(toDegrees): ditto
> 
> Index: java/lang/Math.java
> ===================================================================
> RCS file: /cvs/gcc/gcc/libjava/java/lang/Math.java,v
> retrieving revision 1.6
> diff -u -r1.6 Math.java
> --- java/lang/Math.java 26 Aug 2003 23:14:07 -0000 1.6
> +++ java/lang/Math.java 8 Oct 2003 18:57:42 -0000
...
> @@ -624,7 +630,7 @@
>     */
>    public static double toRadians(double degrees)
>    {
> -    return degrees * (PI / 180);
> +    return (degrees * PI) / 180;
>    }
>  
>    /**
> @@ -638,6 +644,6 @@
>     */
>    public static double toDegrees(double rads)
>    {
> -    return rads * (180 / PI);
> +    return (rads * 180) / PI;
>    }
>  }
> Index: java/lang/StrictMath.java
> ===================================================================
> RCS file: /cvs/gcc/gcc/libjava/java/lang/StrictMath.java,v
> retrieving revision 1.2
> diff -u -r1.2 StrictMath.java
> --- java/lang/StrictMath.java 26 Aug 2003 23:14:07 -0000 1.2
> +++ java/lang/StrictMath.java 8 Oct 2003 18:57:43 -0000
> @@ -1213,7 +1213,7 @@
>     */
>    public static double toRadians(double degrees)
>    {
> -    return degrees * (PI / 180);
> +    return (degrees * PI) / 180;
>    }
>  
>    /**
> @@ -1226,7 +1226,7 @@
>     */
>    public static double toDegrees(double rads)
>    {
> -    return rads * (180 / PI);
> +    return (rads * 180) / PI;
>    }
>  
>    /**
> 



More information about the Java-patches mailing list