Hi Martin,
On 02/07/2013 10:29 AM, Martin Buchholz wrote:
On Thu, Feb 7, 2013 at 9:49 AM, Alan Bateman <alan.bate...@oracle.com>wrote:
On 07/02/2013 17:40, Martin Buchholz wrote:
Alternatively, maybe this is a job for an annotation processor.
IIRC, we considered that 6 years ago. It might be possible, but consider
that you need to diff multiple jdk implementations, and the annotation
processor needs to be able to grok javadoc comments. Also, there are no
"annotations" being processed, so it's weird to use an annotation
processor. Writing a javadoc doclet is another approach that might work.
IIRC, your @since checker was based on parsing javadoc output. If so,
it (like BlenderRev) might have been rendered unusable out of the box
due to the changes in javadoc output made since those tools were written.
Annotation processors don't need to process annotations, they are
general-purpose meta-programming tools that happen to look at
annotations :-)
For such a tool to work, it needs to be able to know what was *really*
part of JDK (N-1), regardless of @since mark-up, to find out what is now
being added in JDK N.
I strongly agree that this property should be checked by some kind of
tool since engineers have a long history of forgetting to add the tags
at the right time.
-Joe
