jkesselm 00/12/15 13:17:29
Modified: java/src/org/apache/xpath DOMHelper.java
Log:
Some doc fixes, some open questions
Revision Changes Path
1.19 +20 -6 xml-xalan/java/src/org/apache/xpath/DOMHelper.java
Index: DOMHelper.java
===================================================================
RCS file: /home/cvs/xml-xalan/java/src/org/apache/xpath/DOMHelper.java,v
retrieving revision 1.18
retrieving revision 1.19
diff -u -r1.18 -r1.19
--- DOMHelper.java 2000/12/15 20:00:44 1.18
+++ DOMHelper.java 2000/12/15 21:17:28 1.19
@@ -332,12 +332,17 @@
/**
* Figure out if child2 is after child1 in document order.
+ * <p>
+ * Warning: Some aspects of "document order" are not well defined.
+ * For example, the order of attributes is considered
+ * meaningless in XML, and the order reported by our model will
+ * be consistant for a given invocation but may not
+ * match that of either the source file or the serialized output.
+ *
* @param parent Must be the parent of both child1 and child2.
* @param child1 Must be the child of parent and not equal to child2.
* @param child2 Must be the child of parent and not equal to child1.
- * @returns true if child 2 is after child1 in document order.
- *
- * NEEDSDOC ($objectName$) @return
+ * @return true if child 2 is after child1 in document order.
*/
private static boolean isNodeAfterSibling(Node parent, Node child1,
Node child2)
@@ -367,6 +372,7 @@
int nNodes = children.getLength();
boolean found1 = false, found2 = false;
+ // Count from the start until we find one or the other.
for (int i = 0; i < nNodes; i++)
{
Node child = children.item(i);
@@ -397,9 +403,17 @@
}
else
{
-
- // NodeList children = parent.getChildNodes();
- // int nNodes = children.getLength();
+ // TODO: Check performance of alternate solution:
+ // There are two choices here: Count from the start of
+ // the document until we find one or the other, or count
+ // from one until we find or fail to find the other.
+ // Either can wind up scanning all the siblings in the worst
+ // case, which on a wide document can be a lot of work but
+ // is more typically is a short list.
+ // Scanning from the start involves two tests per iteration,
+ // but it isn't clear that scanning from the middle doesn't
+ // yield more iterations on average.
+ // We should run some testcases.
Node child = parent.getFirstChild();
boolean found1 = false, found2 = false;
