What's new in GPGP2

What's new

• There has been a lot of changes to the GPGP2 design, especially the design of the basic FSM. So most of the things described here are somewhat out-of-date. Unfortunately there is no documentation regarding the recent changes. Consult directly to Regis Vicent vincent@cs.umass.edu for questions/support.

• Support for initiating an FSM via an unknown KQMLMessage is added to the the scripting tool.

• Inside a FSM script: here is an insider's description about the FSM scripting: http://mas.cs.umass.edu/research/gpgp2/insideScript.html.

• A GPGP2 mechanism specification page is here: http://mas.cs.umass.edu/research/gpgp2/gpgp_gist.html. This is purely Decker's 5 mechanisms.

• Ultimately, GPGP2 has to work with the scheduler -- DTC. Make sure you know how to prepare a ttaems for DTC. This is Ping's note about how to do this: http://mas.cs.umass.edu/~pxuan/research/dtc_guide/scheduler.html, which covers things like how to do re-scheduling, how to incorporate non-local information in your ttaems to let the DTC do the work. This may give you some operational understanding of the DTC, which may save you some time in the future.

• Ping took GPGPCoordination off the CVS repository for major changes -- since we are migrating from old TAEMS implementation to the new one, which is completely different libraries, GPGPCoordination.java cannot compile without the old library, which is already delisted from CVS.

• It seems that it would be somewhat less effort if people just use the base FSMCoordination bean and just write a script. To support that Brett, Jonathan, and Ping have made some changes to FSM/FSMCoordination/Script/ and used a somewhat new event model.

• Ping's providing a generic problem solver that is to be used with FSMCoordination and/or Coordination bean(s)...

• Brett is gone for new opportunities in the world and thus Ping is the one to contact if you have problems with the FSM stuff.