I will present work I have done while at MCC and Stanford
Knowledge System Laboratory. This talk exemplifies a methodology
intented to extract theoretically sound requirements for tools
targeted at the early stages of software design.
The first part of the talk presents a set of studies of the
software design process I conducted at MCC. These studies show
that:
(1) The dynamics of early design is opportunistic. (2) Drawings
play a critical role in design conception, not just documentation.
(3) Designers suffer from various problem-solving difficulties,
such as difficulty in simulating and evaluating partial, informal
solutions and difficulty in keeping track of tasks to be completed.
The second part of the talk describes a set of requirements
for early design environments derived from these findings. DesignVision,
a prototype system built from these requirements while I was
at Stanford, will be described briefly. Key requirements of the
environment include respectively:
(1) The (simultaneous) display of any set of software modules
at arbitrary levels of abstraction and from any subsystem. The
unrestricted, smooth navigation between any of these modules.
(2) Multiple design notations ---pictorial and symbolic---
cross-referenced, editable, and maintained consistent across
all views. Integrated views of control flow, data flow, and functional
decomposition.
(3) Automatic completeness and consistency check, with visual
indication and listing of modules with constraint violations.
Visual display of execution paths in the solution.
The research and design methodology exemplified here shows
that one can effectively derive useful requirements from user
studies, even for tasks that are still poorly understood.
REFERENCES:
Guindon, R. (1992). Requirements and Design of DesignVision,
An Object-Oriented Graphical Interface to an Intelligent Software
Design Assistant. ACM CHI' 92 Conference Proceedings. Monterey,
CA, May 92.
Guindon, R. (1990). Designing the design process: Exploiting
opportunistic thoughts. Human-Computer Interaction, 5, 305-344.
Guindon, R. (1990). Knowledge exploited by experts during
software system design. International Journal of Man-Machine
Studies, 33, 279-304.