My previous patch for this method was wrong. I amended the Mauve test and committed
the following (which fixes the Mauve test and a failure in the Intel test suite):
2006-07-14 David Gilbert <[EMAIL PROTECTED]>
* java/awt/image/MultiPixelPackedSampleModel.java
(createSubsetSampleModel): Restored argument check, but let null
through.
Regards,
Dave
Index: java/awt/image/MultiPixelPackedSampleModel.java
===================================================================
RCS file:
/sources/classpath/classpath/java/awt/image/MultiPixelPackedSampleModel.java,v
retrieving revision 1.8
diff -u -r1.8 MultiPixelPackedSampleModel.java
--- java/awt/image/MultiPixelPackedSampleModel.java 14 Jul 2006 10:47:51
-0000 1.8
+++ java/awt/image/MultiPixelPackedSampleModel.java 14 Jul 2006 11:00:02
-0000
@@ -194,10 +194,18 @@
* a new instance of <code>MultiPixelPackedSampleModel</code>, with the same
* attributes as this instance.
*
- * @param bands ignored.
+ * @param bands the bands to include in the subset (this is ignored, except
+ * that if it is non-<code>null</code> a check is made to ensure that the
+ * array length is equal to <code>1</code>).
+ *
+ * @throws RasterFormatException if <code>bands</code> is not
+ * <code>null</code> and <code>bands.length != 1</code>.
*/
public SampleModel createSubsetSampleModel(int[] bands)
{
+ if (bands != null && bands.length != 1)
+ throw new RasterFormatException("MultiPixelPackedSampleModel only"
+ + " supports one band");
return new MultiPixelPackedSampleModel(dataType, width, height,
numberOfBits, scanlineStride, dataBitOffset);
}
