3.1 Problems of Contemporary Modelling 81
has become true: although many novel methods, tools and improvements were
introduced in software engineering, none of them was a panacea (a universal and
perfect solution) to the existing problems and difficulties. In particular, the
accidental problems of software modelling are tightly bound nowadays to the three
branches mentioned above – software engineering, hardware engineering and
informatics.
3.1.1.3 System Centricity of the Conventional (Approach to) Modelling
At the beginning of the 21
st
century, the modelling of products and processes is
still performed mostly by means of computer-aided systems, primarily for
historical reasons. The trend began with the spread of formal programming
languages (which corresponds approximately to the second stage in Figure 3.1) in
the late 1950s and especially with the idea of writing each new routine in a way
allowing for its reuse (
cf. the third stage in Figure 3.1). Such parameterized
routines were collected into libraries and each newly written program was able to
simply call them and rely on their functionality. A higher number of routines in the
library used to mean higher functionality, therefore there was a naive striving to
make the libraries larger and to write reusable programs. With the growth of
routine diversity in the libraries, one began to lose track of what was available, so a
new trend emerged for reorganization of the routines and grouping them in
purpose-oriented libraries. Then a new type of programs was due, that would allow
use of as many of the library routines as possible with as little effort as possible.
These new program types were initially called applications (or application
software). The latter were also known under different domain- and purpose-
dependent names (
e.g., word processor, spread-sheet, plotting program, etc.), but
were actually the first computer aided software systems, known under the short
name
CAx-systems.
With time, the use of CAx-systems as modelling tools became the dominant
approach to development of complex models – at least in the field of mechanical
and electronic engineering. We shall call this approach the
system centred
approach (SCA), or conventional approach to modelling. For about a decade or
two (say, between 1975 and 1995) the CAx-systems seemed irreplaceable and
untouchable aid in the engineering processes, and especially for engineering
modelling. However, the continuously increasing number of types of CAx-systems
(continuing even nowadays) started to produce software systems deviating from
the “typical CAx-system”–
i.e. distributed systems, intelligent systems, etc. A
common characteristic of all CAx-systems, though, remains the permanently
increasing number of suppliers that are involved in their production. This situation
is relatively easy to explain: permanently increasing requirements lead to higher
required functionality, which, in turn, means involvement of more and more
domains. Consequently, teams developing a given CAx-system involve more and
more experts, which – due to specialization – (have to) come from more and more
different enterprises.
It is highly probable that in the future the focus will be shifted from CAx-
systems, which are actually only tools helping us to build models, towards the
models/results themselves, which are what we need. Thus, it is not impossible that
the time of The System comprised of many autonomous intelligent models (
cf. the
last stage in Figure 3.1) will become reality. Until then, the SCA remains a