🚀 go-pugleaf

hwrvo@kato.lahabra.chevron.com (W.R. Volz) writes:
: It seems that a good design technique should converge much faster than that.
: Perhaps the problem is that there are too many degrees of freedom leading to
: a large solution space. I don't know. What I do know is that it takes too 
: long to design something.

:Some design is needed but spending 90% of the budget on
:the design phase is not right.

I think the problem is the notion that somehow you have to be able
to come up with The Correct Design from the beginning using pure thought alone
and never *dare* touch a keyboard or write a program.

This is also a hopeless optimization procedure, if you only get one
shot at evaluating the optimization function.  (And then have the gall to
say, of course it's right because we Designed it).

In my line of work, I *never* know ahead of time what the program will
eventually until I see the results of the earlier versions.

Maybe what we need is to learn and teach a style of programming that lets
one undo "mistakes" without causing undue damage.  Languages and procedures
that let you do that will of course permit the inevitable modification and
maintenance.

: Bill Volz
: Chevron Oil Field Research Co.
: Exploration Research/Geophysics Division.

--
-Matt Kennel  		mbk@inls1.ucsd.edu
-Institute for Nonlinear Science, University of California, San Diego
-*** AD: Archive for nonlinear dynamics papers & programs: FTP to
-***     lyapunov.ucsd.edu, username "anonymous".