We select programming-in-the-large as our perspective. We do not
try to cover the implications of this statement in extension, and
therefore do not need a definition. We shall, on the contrary, focus
on two aspects.
We take for granted that any significant system, beyond a certain
limit of complexity, is inherently distributed, and shall rapidly give
a few reasons.
We shall then briefly review why incremental development may be
considered appropriate, although we cannot argue much deeper in favour
of this choice with respect to other possible ones.
The second part of this paper will be devoted to identify and
address issues raised while modifying the distributed configuration of
a system, during the incremental development.