On Jan 6, 2009, at 6:19 AM, [email protected] wrote:
Author: jlaskowski
Date: Tue Jan 6 06:19:28 2009
New Revision: 731945
[...]
+ for (int pos = 0; pos < tempClassInfos.size(); pos++) {
+ ClassInfo classInfo = tempClassInfos.get(pos);
+ try {
+ String superType = classInfo.getSuperType();
+ for (Class clazz : classes) {
+ if (superType.equals(clazz.getName())) {
+ classes.add(classInfo.get());
+ tempClassInfos.remove(pos);
+ annClassFound = true;
+ break;
+ }
+ }
+ } catch (ClassNotFoundException e) {
+ classesNotLoaded.add(classInfo.getName());
+ } catch (NoClassDefFoundError e) {
+ classesNotLoaded.add(classInfo.getName());
+ }
+ }
We should change up this algorithm to either not remove while
iterating (i.e. maybe instead add to another list that starts out
empty) or add a "pos--;" after the remove statement. Basically, when
you remove an item the next item becomes the current value of
position, but the loop's next iteration goes to position +1 resulting
in the next item being skipped on every remove. Here's an example:
public class FooTest extends TestCase {
public void test() throws Exception {
List<Integer> list = new
ArrayList(Arrays.asList(1,2,3,4,5,6,7,8,9));
for (int i = 0; i < list.size(); i++) {
list.remove(i);
}
for (Integer item : list) {
System.out.println(item);
}
}
}
The result of this is that all the even numbers will still be there.
Looking at the code though, I bet if we changed classInfos from a list
to a map it would make that whole algorithm a little easier to write.
Optionally, we could even add a "List<ClassInfo> subclasses" and
"ClassInfo superclass" to the ClassInfo and link them together in the
related ClassFinder constructors.
It's great to have you in the code!
-David
