W. Morven Gentleman

Mixed language programming

Subroutines for numerical computation have in the past been written mainly in Fortran and in Algol 60, whereas most programming in the future is expected to be in more advanced languages, for example in Ada. It seems to be a great waste if it will become necessary to convert all existing numerical software into the new languages. What is needed is a facility to include subroutines written in the old languages into programs being written in the newer languages.

Effective use of COTS (commercial-off-the-shelf) software components in long lived systems (tutorial)

This tutorial looks at kinds of COTS software components that can be used in long lived systems, and the technology available for building around them. The potential benefits and risks of this approach to systems are examined. Modifications of conventional development processes are required to focus on where time and cost expenditures occur, and where risks arise.

Administrators and multiprocessor rendezvous mechanisms

This paper discusses rendezvous on multiprocessors. Three different approaches are compared, represented by three specific systems: Ada, Harmony and BNR Pascal. All three permit tasks to run on multiple processors and use blocking communications primitives, but there are significant differences. For example, control over replying to messages out of sequence and over the allocation of tasks to processors is omitted in Ada, but is available in Harmony. The approach represented by BNR Pascal follows a middle road between Harmony and Ada: a low level protocol, invisible to the programmer, is used to ensure communications reliability, but the programmer is aware of when a rendezvous is remote. If performance considerations and verbosity and robustness are ignored, all three approaches are equivalent. To illustrate this equivalence, and to demonstrate clearly the complexity of the Ada rendezvous, an Ada rendezvous administrator written using Harmony is described. A second method of adapting Harmony to Ada is also presented, in which the Harmony primitives are modified to be closer to Ada. In practice, using Harmony primitives directly will usually result in better programs. It is argued that something very much like the rendezvous adminstrator is needed for any actual implementation of the Ada rendezvous.

Hardware assisted high-level debugging

Hardware assistance has long been used for logic level and functional unit level hardware debugging, as well as for machine language level software debugging. Such hardware assistance includes probes to detect signals, comparators to identify matches with expected patterns, buffers to record selected events, and independent logic and software to analyze and interpret the observed events. It can also include the ability to generate selected signals to stimulate the object being debugged and the ability to isolate it from normal changes so its state can be examined. Through knowledge of the data structures and algorithms used by the operating systems, and the runtime representation, register usage, and code bursts produced by compilers, it is possible to take advantage of such hardware assistance in high-level debugging. High-level debugging here refers to debugging in terms of abstractions supported by the operating system and programming languages, as well as user defined abstractions built on top of these. This paper discusses design considerations behind a project to build such a hardware assisted high-level debugger.

Architecture for Software Construction by Unrelated Developers

Suppose one COTS (Commercial Off the Shelf) software supplier provides an interpreter for a problem oriented language, another provides an application generator for producing numerical solvers for a class of partial differential equations, and a third produces a visualization package. A team of domain specialists writes scripts in the problem-oriented language to define cases to be solved, uses the application generator to produce an appropriate solver, solves the generated PDE, and uses the visualization package to analyze the results and adjust the description of cases. Such examples illustrate that large and long-lived software systems can result from the combined efforts of various unrelated development organizations, organizations not even known to one another. No single design authority, to which the others report, has overall system responsibility. Such examples also illustrate the importance of including in software architecture the relationships between entities that exist and are used during the construction process, instead of focusing only on relationships between entities that exist at run time. The needs for software architecture for such systems are not well met by the existing literature.

Mass market computers for software development

W. Morven Gentleman and Mardi Wein Software Engineering Laboratory, Institute for Information Technology National Research Council of Canada Ottawa Ontario Canada KIA OR6 <surname> @iit. nrc. ca The mass market or applitmce computer lends itself well to being used for software development and leads to a different style of development. By being truly personal, it provides a focus for all professional activities of a developer. The primary form of software used on such computers is shrink wrapped software, which has profound implications how the software is used and causes the make-or-buy trade-off to take on an entirely new meaning. This environment, consisting of a network of mass market computers, has infhsenced the development of a novel source management scheme, a summary of which is given here.

COTS and military CCIS

The emergence over the past two decades of an active marketplace of commercial software products has dramatically changed the way computers are used. Many government information system, however, and especially such military systems as Command and Control Information Systems, have not been a big enough market to attract suppliers of commodity commercial software products, and indeed often have characteristics somewhat different from the general marketplace. COTS-based systems represent a way whereby the more specialized systems needed by the military can take whatever advantage they can from Commercial-Off-The-Shelf (COTS) software products.

CASE STUDIES OF REAL-TIME PROCESSING IN ROBOTICS*

Examples are used to illustrate how robotics is computationally demanding, particularly with respect to time constraints within which the computation must complete. The examples are picked both from control algorithms for robots and from sensor data interpretation for robotics. The linear algebra problems involved are indicated.

Parallel Computing You Can Do

Until a few years ago, parallel computing was an esoteric subject of interest only to theorists or to the few pockets of experimental computer science with real machines. Everyone knew that ultimately, for any given hardware technology, the only way to increase speed is to go to multiple processors, but such machines were rare, expensive, and complicated to use. Generally, the search for speed was satisfied by faster hardware — the only form of parallelism to make significant impact from the mid 70’s to the mid 80’s was vector computation. What has changed in the past few years is that microprocessors, and to a lesser extent VLSI, have made building parallel computers easy and inexpensive. Many architectures have been implemented, and manufacturers have found that customers will buy machines, if the machines are not too expensive, just to familiarize themselves with parallel computing, to experiment with how parallelism can be used in their problems, and to compare different ways of using parallelism. The availability of real machines has caused a radical shift in the computing community. Parallelism has become fashionable. There have been many conferences and journal issues on the subject. Even the general news media report on progress. The activity has spurred algorithm development and theoretical analysis. Myths and misconceptions have been debunked. There have been spectacular successes. This spring, the Bell and Karp prizes were awarded to a team from Sandia who used a 1024 processor hypercube, a general purpose computer, to obtain speedups of more than 500 on three diverse physical problems over what a single processor would have taken on those problems — which they prefer to describe as a speedup of 1020 over the largest versions of these problems that a single processor would have been used to solve.