Future Astronomical Systems

Ted Groner (ted@sparky.sao.arizona.edu)
Fri, 20 Oct 1995 07:00:12 +0100

To Gentle Persons of the Software Persuasion:

Having been a major "lurker" so far on the fads discussion, and wishing
to climb onto the last-minute post bandwagon, let me fill yer ear!

I think the concept of interoperability is a superb idea. (see GNU and
Linux) Perhaps the practicality is another matter. The task is daunting,
and the darkest hours last forever. Use the force, Luke.

Major players in the astronomical field, such as iraf, IDL, MIDAS, etc...
contain little interoperability. Minor players such as myself have never
seriously considered this idea. All too often it is exceedingly hard to
get only five tightly grouped programmers to agree the sun will rise
SOMEWHERE in the morning! And don't you dare add even one astronomer!
He would attempt to reprogram the computer controller of his car while
attempting to pass a timber truck on a 7% uphill grade in the Sierra's
on snow, just to see if it would reduce the rattle in the passenger side
mirror!

What could be of greatest benefit to all however IS a standard. If we
can agree to standard that our group alone would attempt to use when at
all possible, we have accomplished a major task.

The detail of scripting languages, OOP or not to OOP, etc... should
follow a discussion about the function/procedure call level of interface.
The overlay of window systems, varying hardware, user community divergence,
fill in your own blank, make this extremely unlikely.

However, the unlikely does on occasion occur. I am not suggestion we
forget discussion now or in the future on these topics. A more united
front against common problems can help us all. Let us first attempt to
address a very basic question. If, indeed we can agree to write code
usable to our own community, we must first agree to basics. Perhaps
a standard for shared C routines, and or C++. Fortran is still used
widely. I myself would follow the GNU path, and use C in gcc dialect,
since it has wide interoperability.

Once we all can use a common package of low level routines, higher end
user interface issues should become easier to resolve. Quite a bit of
software exists freely that most of us access today. Many packages are
available that produce linkable libraries. It does seem they lack a common
interface at the programming level. Let us choose one!

See you there!
Ted

Help Stamp Out (|:-}+-|.....||||||.....||--===| (Programmer-Truck Rollovers)
Ted Groner Smithsonian Astrophysical Observatory, FLWO
ted@sparky.sao.arizona.edu (520) 670-6746