[Stellar-discuss] a software framework

[Piet Hut]

Dear MODEST colleagues,

Greetings from Lund, Sweden, where we just concluded the workshop
MODEST-6a, organized successfully and efficiently by Melvyn Davies:
"http://www.astro.lu.se/~melvyn/modest6a.html" .

One of the many highlights of the meeting was a discussion about
software frameworks, centered around the question of interoperability
of software tools.

Many of us have written tools that we could share with others,
but in practice we often wind up rewriting what others have done,
because of a lack of ease in mixing and matching different tools.
Most software tools have little or no documentation, and often
each tool has its own input/output format and ideosyncratic way
of addressing its options through command line argument or
configuration file.  A central question is: how to overcome these
obstacles without insisting on straitjackets; in other words: how
to let everyone free to write in their own way and style, and yet
allow an easy sharing of software modules.

During the workshop, we discussed the advantage of developing a
software framework within which it will become easy to couple
different tools together.  We will not insist on any particular
form or format or language or whatever other aspect of the innards
of each tool.  The idea is that all that will be hidden in a black
box, with only the wrapper interface visible to the typical user.

The main challenge will be to define the interfaces and data formats,
while starting off with a series of toy models, almost trivial black
box examples that do little or nothing but have at least the right
interface to be hooked up with other such modules.  From there on,
we can gradually add more and more astrophysics to the toy models,
to let them "grow up" to become real astrophysics tools.

Ross Church has kindly set up a special email list to discuss this
tool building process.  You can subscribe by going to the MODEST
web site  "http://www.manybody.org/modest.html"  and click on
"mailing list".  You will then find, in the last sentence, a
"here" to click on to bring you to Ross's stellar-discuss page.

Briefly, the stellar-discuss email list is aimed at discussing
how to create and connect software tools for anything stellar,
from stellar evolution to stellar dynamics to stellar hydrodynamics
and other stellar astrophysics, within the context of modeling dense
stellar systems.

[historical note: this list was set up a few weeks ago by Ross in
 response to an initiative by Bill Paxton, who had polled a few
 people working on stellar evolution codes.  A big hand for Bill
 and Ross for creating a play pen for us to work in!]

Please come join us, if you like to plug 'n play, in a MODEST way,

Piet