Workshop minutes on the Jini Pattern Language

• Direction of Jini - Jini is going to evolve from where it is now. As Jim Waldo explained, some of the new directions may be:
  • Surrogate proxies
  • Jini and EJB interoperation
  • Further integration of JavaSpaces
  • Searching for a simple, yet powerful, security mechanism

• Discovered patterns - Many of the presentations aimed at finding patterns in Jini and ad hoc networking in general. Presentations, which were mainly focusing on patterns are:
  • Event Pipeline Pattern, by Bina Ramamurthy
  • Pick and Verify Pattern, discovered in UPnP, by Markus Wischy
  • Service-Oriented Programming Pattern Language, by Guy Bieber
  • Locate and Track Pattern, by Bruce Cohen
  • Profile-based Service Browsing, by Martin Gitsels
  • User Interface Pattern, by Vrijendra Gokhale

• Jini and Components - Jim Waldo scratched the topic with the EJB/Jini integration slightly. It got covered in detail by Guy Biebers presentation about the OpenWings project. During that presentation, it became clear that service-oriented programming, which he is suggesting, bases on component - container contracts. A component is the deployment unit of a service.
  Also the JinACE presentation of the workshop organizers has been pointing into that direction. However the focus here was not restricted to Java, but to ad hoc networking in general. Some ideas include making the CORBA Component Model (CCM) ad hoc capable by merging it with dynamic scenarios from the Jini world. The other direction, of making todays ad hoc networking frameworks, like Jini, more component-oriented, by introducing techniques like the deployment and configuration scenarios, from EJB and CCM.

• Location awareness - in ad hoc networking seems to be one of the big problems and therefor an important field of research. Fabio Kon raised some issues, problems in this field. With the "2k Name Service Browser" he showed how they tackled the problem in their environment.
  Bruce Cohen aimed at the same problem with his Locate and Track pattern, where he suggests to make the location one of the criterias of the service discovery.
  The "Profile-based Service Browsing" presentation aimed at a similar direction, where services would register with a profile, possibly containing location information, which gets then used for enhanced searches on the services.
  Location awareness can be seen as one of several parameters found by reflection on the environment the service is running in.

• Scaling down - The currently available frameworks for ad hoc networking are too big to fit on todays small devices. Jim Waldo and Fabio Kon mentioned some ideas around that theme.
  • Surrogate architecture in Jini
  • Proxies at various levels responsible for gateway tasks
  • Bridges between technologies, e.g. Jini/CORBA bridge

• Suggested areas of work - Due to the complexity of the themes and the restricted time we left the workshop with many open questions. Here a list of them:
  • How to standardize location awareness?
  • Discover more patterns, e.g. in the areas of Security, Proxies, Bridges and Federations.
  • Find a simple but efficient security mechanism appropriate for small devices
  • Simplicty - ad hoc networking frameworks should be simple in the core and extensible with services around them.
  • How can loaded services be replaced? Jini specifies how services get loaded, but how do they get unloaded?
  • Relection techniques might be of matter to specific domains of small devices. How can such information be accessed in a standard way?