On 10/25/2013 12:29 PM, Peter Levart wrote:
Hi Joe,
So, the order must be respected, right.
There's a bug. I think you wanted to call:
((Class) this).getSuperclass()
Instead of:
thisClass.getSuperclass()
...which would always return Object.class...
Well, this is not the only change needed, of course. What about
something like the following (prefer iteration to recursion and use
AnnotationType.isInherited() which is faster):
default <T extends Annotation> T[] getAnnotationsByType(Class<T>
annotationClass) {
T[] result = getDeclaredAnnotationsByType(annotationClass);
if (result.length == 0 && this instanceof Class &&
AnnotationType.getInstance(annotationClass).isInherited()) {
for (Class<?> clazz = ((Class<?>) this).getSuperclass();
clazz != null; clazz = clazz.getSuperclass()) {
result =
clazz.getDeclaredAnnotationsByType(annotationClass);
if (result.length > 0) {
break;
}
}
}
return result;
}
Regards, Peter
Regards, Peter
On Oct 25, 2013 10:40 AM, "Joe Darcy" <joe.da...@oracle.com
<mailto:joe.da...@oracle.com>> wrote:
Hi Joel and Peter,
On 10/24/2013 07:10 AM, Peter Levart wrote:
Hi Joe,
I see two problems with the implementation in
*AnnotatedElementSupport*. The first is the treatment of
declared-only annotations where the code returns either directly
present or in-directly present repeatable annotations, but not
both. So in the following example:
@Ann(1)
@AnnCont({@Ann(2), @Ann(3)})
it will only return [@Ann(1)], but I think it should return all
of them [@Ann(1), @Ann(2), @Ann(3)] - does the spec. define that?
[snip]
From your feedback (and a closer reading of the specifciation),
I've reworked the specifications and implemenations of the default
methods for get[Declared]AnnotationsByType:
http://cr.openjdk.java.net/~darcy/8005294.2/
<http://cr.openjdk.java.net/%7Edarcy/8005294.2/>
Tests still need to be written, but this implementation should be
much closed to what is needed.
Thanks,
-Joe
