Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision		Previous revision
research:mao:new-things [2010/11/24 15:51] – created etanterresearch:mao:new-things [2010/11/24 17:32] (current) – etanter
Line 1: Line 1:
 +====== New things! ======
  
 +shared library (potential exhibit) + "wide" quantification (dynamic-analysis-like)
 +
 +<code>
 +xpi Jps {
 +  joinpointtype Object Creations(Class c);
 +  
 +  // pointcut library of "meaningful pointcuts for the JPT"
 +  pointcut creations(Class c) : call(* *.new(..)) && target(c);
 +  
 +  // other alternative
 +   joinpointtype Object Creations(Class c) defaults: call(* *.new(..)) && target(c);
 +}
 +
 +module Collections  {
 +  contains java.util.*;
 +  potentially-exhibits Object Jps.Creations(Class c) : creations(c);
 +  //alternative using the default
 +  potentially-exhibits Jps.Creations;
 +}
 +
 +
 +
 +module M1 {
 + contains C;
 +}
 +
 +class C {
 +  void foo() { 
 +     m = new HashMap(); 
 +     m.add(x); // creates objects within Collections
 +     ... 
 +  }
 +}
 +
 +module M2 {
 +  contains C;
 +  exhibits Object Jps.Creations(Class c) : creations(c);
 +  //alternative using the default
 +  exhibits Jps.Creations;
 +}
 +
 +module Aspect {
 +  contains Asp;
 +}
 +
 +aspect Asp {
 +  around Object Jps.Creations(Class c) {
 +    if (shouldBeCached(c)) ...pool...
 +    else return proceed(c);
 +  }
 +}
 +
 +RESULT:
 +Asp does not see HashEntry creations caused by C@M1.
 +Asp sees HashEntry creations caused by C@M2, as well as the HashMap creation in C.
 +</code>
 +
 +symmetric case, beyond levels (topological scoping a la membranes)
 +<code>
 +module Prof {
 +  contains prof.*;
 +  advises Base, Racer;
 +}
 +
 +module Base {
 +  contains ...;
 +  exhibits Jps.Creations(..) : ...
 +  exhibits Jps.FieldAccesses(..) : ...
 +}
 +
 +module Racer {
 +  contains racer.*;
 +  advises Base;
 +  exhibits Jps.Creations(..) : ...
 +}
 +RESULT: Prof sees creations in both base and racer (not feasible with levels without reintroducing loops)
 +</code>
 +
 +now with transitive topological usage
 +
 +<code>
 +module Prof {
 +  advises M1;
 +}
 +
 +module M1 {
 +  contains D (that calls objects in M3)
 +  exhibits X;
 +  uses M3;
 +}
 +
 +module M2 {
 +  contains C (that calls objects in M3)
 +}
 +
 +module M3 {
 +  exhibits X;
 +}
 +RESULT: 
 +Usage of M2 does not trigger X jps visible by Prof. 
 +Instances of X in M3 are generated and seen by Prof only in the cflow of M1. 
 +(removing "uses M3" in M1 would avoid this)
 +</code>