time-line

[cgl]

Just a hint about the timeline of swarm-releases as we currently 
envision them...to be taken with a *small* grain of salt....;-)

We are adding about 25 new sites to the beta program this evening
(Monday, March 4th). 

We are aiming to release a generally available beta version by
March 31 (and not a day later, but possibly 2 days later!!)

Primary goals for this generally available beta version of swarm
are to have the documentation a lot more fleshed out, especially
in the area of the collection and defobj libraries. A number of
demo applications will be added, including one to illustrate
multi-level swarms, and one to demonstrate experiment-swarms.

Method probes will be available in addition to variable probes.

Support for experiment control swarms will be in place, at least in
a basic form, allowing the creation, monitoring, data-analysis, and
filing of multiple runs sweeping over parameter spaces....all through
the same swarm scheduling mechanism you already use to build
worlds and run them - This will probably be built on MPI, so 
that multiple versions of swarm experiments can be managed over
a network of processors.

A bunch more graphic user tools will be in place, hopefully including
the ability to dump and examine the dynamic schedule structure, and
an interactive editor for creating and/or altering schedules on
the fly.

Finally, we expect a couple of GIS applications, with artificial
agents running around on real-world data-bases to be in
place by March 31. Some of the prime-movers behind GRASS
are joing in this release, and have promised to work with
us to make Swarm and Grass more interactive with one another....

Slightly longer term, we are aiming to have V1.0 ready for release
by early June. This will (hopefully!) be pretty stand-alone, and
users should not have to query us very much about how to get
apps up and running. We expect that the period after that, at least
the summer of 96, will be devoted to mutually assisting each other
in bringing our swarm-apps up and running, and producing "real" data.

Hopefully, as people add tools/objects/agents/spaces that they need
in their experimental domain, they will write the tools up in a general
enough purpose, well-documented way, in order that they may be 
contributed back to the Swarm libraries, to make them useful for
others. We are in the final throes of determining the proper format
and content of contributed library objects....help/suggestions
are appreciated.

In the longer run, Manor Askenazi will be building the parallel
version of the Swarm kernel, but it is hard to set a date on
a beta-release for that - next fall at the earliest, probably
at the end of the year....

Cheers!

Chris Langton