name suggestions:

  * x modules: actually, what about **X-Modules**? (ie. "crosscutting modules", nice mix between XPI and Open Modules)
  * aspectual modules:  already (rarely) used
  * aspect(ual) shells
  * aspect(ual) capsules

----

  * [[research:mao:iiia-examples|IIIA Examples / problems]]
  * [[research:mao:advice-dispatch|Advice dispatch]]
  * [[research:mao:subtyping|Subtyping considerations]]
  * [[research:mao:two-signatures|Extended JPI signatures as in StrongAspectJ]]
  * [[research:mao:invariant-pcd|Invariant Pointcut Designators]]
  * [[research:mao:overloading|JPI overloading]]
  * [[research:mao:parametric-polymorphism|Extending JPI with parametric polymorphism]]
  * [[research:mao:extended-signatures|Extending JPI Signatures]]

----

====== General Design Principles/Decisions ======

Following Ungar's grid of Values - Principles - Practices

  * values (what you care about, what you want, what you like)
      * teamwork
      * reuse
      * change
      * simplicity

  * design principles (general rules for achieving values)
      * uniformity/symmetry (S)
      * separate development, separate compilation (TC)
      * loose coupling (RC)
      * modular reasoning (TRC)
      * make it hard to shoot yourself in the foot / avoid surprises (S) [loops, double jps]
      * support for shared libraries (R)
      * coarse-grained quantification (S)
      * abstraction (TRCS)

  * specific practices (specific things to do)
      * modules encompassing classes/packages (cgq, sdsc, mr, lc, ssl)
      * modules can both advise and exhibit (us)
      * join point types (as, mr, lc, sdsc)
      * functional join point types (no structs) (as)
      * explicit declaration of exhibited jpts (as)
      * join point subtypes (a)
      * explicit join points / block jps (a, as)
      * "potentially-exhibit" declarations (shl)
      * compile-time checks for potential loops (as)
      * execution levels / or notion of cflow based on use-relation (as)



----

====== Examples ======

  * [[research:mao:iiia-examples-fixed|fixed versions of IIIA buggy examples]]
  * translate good examples from IIIA, closure jp paper, PPPJ explicit jp paper, open modules paper(s), XPI paper
  * [[research:mao:new-things|new things we can do]] that they can't (dynamic analysis aspects -- wide scope, etc.)
  * case study:
      * Racer / Profiler
      * AJHotDraw
      * Berkeley DB
      * AspectOptima
      * HealthWatcher
      * iBatis
      * MobileMedia

====== Pending ======

  * TODO: each class contained in only one module? Java8: different namespaces/loaders, incompatible
  * Study the Java8 proposal (link?)
  * explicit declaration of used modules (sdsc, lc)  [removed?!]
  * references: XPIs, Open Modules, ECeasarJ/EScala

====== Paper outline ======
modular reasoning / implicit invocation / AOP
  * motivation from IIIA
  * overview of IIIA and limitations/problems
  * overview of X-Modules (model, boxes and arrows)
  * X-Modules in practice: addressing IIIA issues
  * implementation: AspectJ extended / abc, implementation details, separate compilation, checks, etc. 
  * case study
  * related work