[aspectj-users] Fluent interface for Pointcuts in @AspectJ style Aspects

[Thomas Darimont <thomas.darimont@xxxxxx>]

Hello,

I don't know whether this was already discussed on the list (at least I 
couldn't find the thread) so I ask
what do you think of supporting a fluent API for Pointcuts in @AspectJ 
Java Style Aspects?

It would look like this:
/**
*
*/
package de.tutorials.aop.aspectj;

import org.aspectj.lang.ProceedingJoinPoint;
import org.aspectj.lang.annotation.Around;
import org.aspectj.lang.annotation.Aspect;

import de.tutorials.aop.Any;
import de.tutorials.aop.Arguments;
import de.tutorials.aop.NotNull;
import de.tutorials.aop.Pointcut;
import de.tutorials.aop.Pointcuts;
import de.tutorials.aop.Transactional;

/**
* @author Thomas.Darimont
*
*/
@Aspect
public class Retry {

   int maxRetries;

   public int getMaxRetries() {
       return maxRetries;
   }

   public void setMaxRetries(int retryLimit) {
       this.maxRetries = retryLimit;
   }

   public Pointcut someMethods() {
       return Pointcuts
                   .call()
                   .methods("process*")
                   .withinPackage("de.tutorials.services..")
                   .withAttribute(Transactional.class)
                   .withArguments(new Arguments() {
                       Any args; //Any -> special Typ matching all 
types like "*"
                       int param0;
                       @NotNull String param1;
                   }).;
   }

   @Around("someMethods")
   public Object retry(ProceedingJoinPoint thisJoinPoint) {
       Object result = null;
       int currentTry = 0;
       while (currentTry < maxRetries) {
           try {
               result = thisJoinPoint.proceed();
           }catch(Throwable throwable){
               //log ...
               currentTry++;
           }
       }
       return result;
   }
}


the obvious Advantages are:
- Pointcut definition would be refactoring friendly (e.g. renaming of a type
- Pointcut Expressions could be checked by the compiler (to a certain 
degree)
- AspectJ could be used to validate Pointcuts at compiletime (declare 
error!) :)

What do you think about this?

Best regards,
Thomas