Large Scale Porting through Parameterization

The advent of open systems and standards, while beneficial, has not eliminated the difficulty of maintaining and transporting large scale software systems across many varying platforms.

In this paper we discuss the need and criteria for a effective porting strategy, one that allows the rapid and inexpensive retargeting of large scale software systems to many widely varying platforms while not compromising the integrity of that software on any previously supported platform.

"Getting Tigger down", said Eeyore, "and not hurting anybody, Keep those two ideas in your head, Piglet, and you'll be all right."
A. A. Milne, The World of Pooh, 1957, pp216, McCelland & Stewart Ltd.

A key component of any porting strategy is the methodology used to determine, represent, use, and validate specifications of the target system's characteristics and site or system dependent build and run time controls. The standards efforts (e.g., POSIX, ANSI C) are attempting to eliminate the large number of discrepancies that exist among systems today. However, the problem will always exist, for reasons that are discussed.

Hence, the main objective of this paper is to present and justify the methodology that we use.

This methodology is in production use on several commercial products in Sietec. Its benefits include relieving the programmer from the burden of needing detailed knowledge of all the idiosyncrasies of the various target environments. It is sufficiently powerful that it accommodates many flavours of BSD, System V, and DOS