MegaCads developers of DLR:
The initial idea of an interactive, parametric grid generation
software tool was developed by Arno Ronzheimern 1992, who also
wrote the initial FORTRAN 77 version of MegaCads (V 1.0).
This version made use of some specific compiler options, and its
use was thus restricted to two hardware platforms
(IBM R6000 and HP 7xx).
All data transfer was based on files which made it real slow,
even with smaller data sets.
The available graphics system was very limited and offered no
possibilities to create a user friendly graphical user interface (GUI).
In spite of the fact, that its memory layout was originally intended
for 2D applications, it had some (limited) 3D capabilities.
It became obvious, that it was necessary to do a complete
redesign of the software, when a full 3D implementation was
becoming necessary.
To develop the new generation of MegaCads, Olaf Brodersen and
Martin Hepperle joined the team. Both had been working with the
first version of MegaCads for several projects and knew its
strengths and shortcomings.
Olaf Brodersen had been working on grids for the solution
of the Navier Stokes equations for the flow around aircraft
configurations, Martin Hepperle had been working on
airfoils, propellers and engine nacelles and brought a vast
programming experience with him.
After lots of brainstorming the path for future development
was becoming clear.
The code of MegaCads 1.0, written in FORTRAN 77,
had to do all its memory management (if you want to call
fixed length arrays 'management') by himself, had serious
shortcomings, its user interface was very rudimentary and
its internal data structures were cryptic.
The first idea, favoured by those old FORTRAN programmers,
was to create
a graphical user interface using the 'C' language and connect
this front end to the existing working routines in FORTRAN.
While this was possible in principle, the combination of
FORTRAN and C has to be handled differently on different
platforms. There is no standard way to call FORTRAN routines
from C routines and vice versa.
This would get even more worse when trying to call C++
functions from FORTRAN.
Similar problems occured, when
a marriage of FORTRAN 90 and C was tried. It became
obvious, that FORTRAN pointers are no pointers at all,
but handles to complex data structures, which vary from
compiler vendor to compiler vendor.
Thus it was finally decided to start over from scratch
and to do all the coding in C.
The future will see the development of more sophisticated support tools
which will (partially) automate the processes of segmentation and
boundary condition specification as well as more sophisticated
geometry primitives (useful for optimization methods).
The application of MegaCads to real world problems will lead to a
set of scripts which will be available for a set of
standard configurations.
Other library scripts will perform the construction of
basic building blocks.
This will speed up the composition of grids for complex configurations.
Additional support through:
The MEGACATS!
All these folks are busy trying to keep away bad news
from your screen:
