The Forgotten Interfaces: A Critique of Component-based Models of Computing

Abstract

Many theoretical methods for dealing with component-based software design have been proposed. Unfortunately, practically all of them ignore the inconvenient fact that software-based systems need an underlying hardware and software infrastructure to function. Consequently, all software applications are susceptible to idiosyncratic effects stemming from the hardware as well as supporting software layers. This layering relationship between software and its supporting hardware and software platform is unprecedented in engineering practice, since it represents a unique blending of the conceptual and physical domains. If we are to build truly reliable component-based software applications, it seems critical that the nature of this relationship is clearly defined and understood. Ignoring it would be irresponsible and, more worryingly, potentially dangerous. In this article, we first analyze the non-trivial nature of this fundamental and unique relationship and also argue what may be the reason why it has been neglected so often. This leads to the notion of engineering correctness as something that is fundamental to reliable software design and which, it is claimed, is inseparable from the traditional and established concept of logical correctness. It is argued further that this requires new engineering-oriented approaches to component-based software application design that must factor the influence of platforms. In the final part of the article, one possible approach for achieving this is described.

Keywords

Component-based Software Development, Verifying and Reasoning about Programs, Formal Languages, The Physics of Software.

Cite as:

Bran Selic, “The Forgotten Interfaces: A Critique of Component-based Models of Computing”, Journal of Object Technology, Volume 19, no. 3 (October 2020), pp. 3:1-8, doi:10.5381/jot.2020.19.3.a2.

This article is accompanied by a video realized by the author(s).