The branches of engineering (such as civil, electrical, and mechanical), are often referred to as disciplines for good reason. Associated with each specialty is a set of rules that specify:
Like all professional education, engineering education is designed to prepare students to meet the requirements of the authorities that regulate their chosen profession. Consequently, most graduates are taught they must carry out these procedures diligently and are warned they can be deemed guilty of negligence and lose the right to practice their profession if they do not.
Because they are preparing students for a career that can last many decades, good engineering programs teach fundamental principles that will be valid and useful at the end of the graduate's career. Engineering procedures are based on science and mathematics; and graduates are expected to understand the reasons for the rules, not just blindly apply them.
These procedures are intended to assure that the engineer's product:
In some areas, for example building and road construction, the procedures are enforced by law. In other areas, and when engineers work in industry rather than selling their services directly to the public, employers rely on the professionalism of their employees. Professional engineers are expected to know what must be done and to follow the rules even when their employer wants them to take inappropriate shortcuts.
Anyone who observes engineers at work knows that exercising due diligence requires a lot of "dog work."The dull, but essential, work begins in the design phase and continues through construction, testing, inspection, commissioning, and maintenance. Licensed engineers are given a unique seal and instructed to use it to signify the acceptability of design documents only after they are sure the required analysis has been completed by qualified persons.
Recent experiences reminded me that the activity we (euphemistically) call software engineering does not come close to deserving a place among the traditional engineering disciplines. Replacing an old computer with a newer model of the same brand revealed many careless design errorserrors that in all likelihood could have been avoided if the developers had followed a disciplined design process. None of the problems was safety critical, but the trouble caused was expensive and annoying for all parties.
My "adventure" began when the sales clerk scanned a bar code to initiate the process of creating a receipt and registering my extended warranty. There were three codes on the box; not surprisingly, the sales clerk scanned the wrong one. This is a common occurrence. The number scanned bore no resemblance to a computer serial number but was accepted by the software without any warning to the clerk. The nonsense number was duly printed as the serial number on my receipt. My extended warranty was registered to a nonexistent product. I was billed, and no problem was noted until I phoned the customer care line with a question. When I read the serial number from the receipt, I was told that I had purchased nothing and was not entitled to ask questions. After I found the correct number on the box, I was told that my computer was not yet in their system although a week had passed since the sale.
Computer science students are not taught to work in disciplined ways. In fact, the importance of disciplined analysis is hardly mentioned
Correcting the problem required a trip back to the store and tricking the company computer by returning the nonexistent machine and buying it again. In the process, my name was entered incorrectly and I was unable to access the warranty information online. After repeatedly trying to correct their records, the help staff told me it could not be done.
A different problem arose when I used the migration assistant supplied with the new computer to transfer my data and programs to the new machine. Although the description of the migration assistant clearly states that incompatible applications will be moved to a special directory rather than installed, a common software package on the old machine, one that was not usable or needed on the new one, was installed anyway. A process began to consume CPU time at a high rate. Stopping that process required searching the Internet to find an installer for the obsolete product.
The next problem was an error message informing me that a device was connected to a USB 1.1 port and advising me to move it to a USB 2.0 port. My new computer did not have any 1.1 ports so I called the "care" line for advice. They had no list of error messages and could not guess, or find out, which application or component of their software would issue such a message or under what conditions it should be issued. They referred the problem to developers; I am still waiting for a return call.
These incidents are so petty and so commonplace that readers must wonder why I write about them. It is precisely because such events are commonplace, and so indicative of lack of discipline, that such stories should concern anyone who uses or creates software.
As early as the late 1950s, some compilers came with a complete list of error messages and descriptions of the conditions that caused them. Today, such lists cannot be found. Often, when reviewing a system, I will pick a random message or output symbol and ask, "When does that happen?" I never get a satisfactory answer.
There are methods of design and documentation that facilitate checking that a programmer has considered all possible cases (including such undesired events as incorrect input or the need to correct an earlier transaction) and provided appropriate mechanisms for responding to them. When such methods are used, people find serious errors in software that has been tested and used for years. When I talk or write about such methods, I am often told by colleagues, experienced students, and reviewers that, "Nobody does that." They are rightthat's the problem!
Much of the fault lies with our teaching. Computer science students are not taught to work in disciplined ways. In fact, the importance of disciplined analysis is hardly mentioned. Of course, just telling students to be diligent is not enough. We need to:
Many of us preach about the importance of determining the requirements a software product must satisfy, but we do not show students how to organize their work so they can systematically produce a requirements specification that removes all user-visible choices from the province of the programmer.
Some of us advise students to avoid dull work by automating it, but do not explain that this does not relieve an engineer of the responsibility to be sure the work was done correctly.
It has become modish to talk about teaching creativity and innovation. We need to tell students that inventiveness is not a substitute for disciplined attention to the little details that make the difference between a product we like and a product we curse. Students need to be told how to create and use checklists more than they need to hear about the importance of creativity.
Even sophisticated and experienced purchasers do not demand the documentation that would be evidence of disciplined design and testing.
It is obviously important to give courses on picking the most efficient algorithms and to make sure that students graduate prepared to understand current technology and use new technology as it comes along, but neither substitutes for teaching them to be disciplined developers.
Disciplined design is both teachable and doable. It requires the use of the most basic logic, nothing as fancy as temporal logic or any of the best-known formal methods. Simple procedures can be remarkably effective at finding flaws and improving trustworthiness. Unfortunately, they are time-consuming and most decidedly not done by senior colleagues and competitors.
Disciplined software design requires three steps:
Each of these steps requires careful review:
These rules seem simple, but reality complicates them:
Similar "divide and conquer" approaches are available for inspection and testing.
While our failure to teach students to work in disciplined ways is the primary problem, the low standards of purchasers are also a contributing factor. We accept the many bugs we find when a product is first delivered, and the need for frequent error-correcting updates, as inevitable. Even sophisticated and experienced purchasers do not demand the documentation that would be evidence of disciplined design and testing.
We are caught in a catch-22 situation:
The Digital Library is published by the Association for Computing Machinery. Copyright © 2010 ACM, Inc.
The following letter was published in the Letters to the Editor in the March 2011 CACM (http://cacm.acm.org/magazines/2011/3/105325).
As a programmer for the past 40 years, I wholeheartedly support David L. Parnas's Viewpoint "Risks of Undisciplined Development" (Oct. 2010) concerning the lack of discipline in programming projects. We could be sitting on a time bomb and should take immediate action to prevent potential catastrophic consequences of the carelessness of software professionals. I agree with Parnas that undisciplined software development must be curbed.
I began with structured programming and moved on to objects and now to Web programming and find that software is a mess today. When I travel on a plane, I hope its embedded software does not execute some untested loop in some exotic function never previously recognized or documented. When I conduct an online banking transaction, I likewise hope nothing goes wrong.
See the Web site "Software Horror Stories" (http://www.cs.tau.ac.il/~nachumd/horror.html) showing why the facts can no longer be ignored. Moreover, certification standards like CMMI do not work. I have been part of CMMI-certification drives and find that real software-development processes have no relation to what is ultimately certified. Software development in real life starts with ambiguous specifications. When a project is initiated and otherwise unrelated employees assembled into a team, the project manager creates a process template and fills it with virtual data for the quality-assurance review. But the actual development is an uncontrolled process, where programs are assembled from random collections of code available online, often taken verbatim from earlier projects.
Most software winds up with an unmanageable set of bugs, a scenario repeated in almost 80% of the projects I've seen. In them, software for dropped projects might be revived, fixed by a new generation of coders, and deployed in new computer systems and business applications ultimately delivered to everyday users.
Software developers must ensure their code puts no lives at risk and enforce a licensing program for all software developers. Proof of professional discipline and competency must be provided before they are allowed to write, modify, or patch any software to be used by the public.
As suggested by Parnas,(1)(2) software should be viewed as a professional engineering discipline. Science is limited to creating and disseminating knowledge. When a task involves creating products for others, it becomes an engineering discipline and must be controlled, as it is in every other engineering profession. Therefore, software-coding standards should be included in penal codes and country laws, as in the ones that guide other engineering, as well as medical, professions. Moreover, software developers should be required to undergo periodic relicensing, perhaps every five or 10 years.
(1) Parnas, D.L. Licensing software engineers in Canada. Commun. ACM 45, 11 (Nov. 2002), 9698.
(2) Parnas, D.L. Software engineering: An unconsummated marriage. Commun. ACM 40, 9 (Sept. 1997), 128.
Displaying 1 comment