Although the clock says there are twenty-four hours in a day, you cannot work twenty-four hours a day. For health’s sake, you need seven to eight hours of sleep a day and about three hours for personal hygiene, meals, etc. It really looks like you can only have fourteen to fifteen hours per day to spend as you want. In a perfect world, you would have 7 × 15 = 105 hours to spend as you like.

Anyone who has been associated with college students knows that students do not spend Friday night, Saturday, or most of Sunday doing school work. And that is as it should be. However, this effectively removes thirty more hours from your time budget: you now have 5 × 15 = 75 hours per week to allocate.

A three-credit university course is really nine hours because you are expected to spend three hours in the classroom and two hours per hour of class preparing for class. The preparation may be reading or homework to hand in or it may be working on a program. Therefore, a fifteen-hour load requires that you spend forty-five hours per week. Although the “standard” work week is forty hours, you can expect to spend a bit more than that each week. Notice, though, that you have seventy-five disposable hours per week.

Unfortunately, most of us cannot be so efficient as to use all the time in-between class. So, let’s say that you spend your days in class (fifteen hours) and that leaves thirty to forty-five hours at night; that’s six to seven hours per night. My experience is that you cannot profitably do that, either. After a long hard day, you are not an efficient user of time. The suggestion is that you find a block of time during some days to get the nighttime load down to something manageable, say four hours a night.

The key to any science is the metrics used to measure activities. The Time Utilization Log (TUL)^* is to be used to record how you’re using your time. It has the class times^† already accounted for between 8:00 a.m. and 5:00 p.m. The rest of the day is blocked off in half-hour increments. The assumption is that you stay up late and get up late. Early risers should modify the grid.

The Time Recording Template (TRT) (note that this form is one of many in the Excel packet) is to be used for recording time spent “programming,” which includes time to design, code, test, and document programs. As with the TUL, the TRT keeps track of your time but the TRT is used to determine how you use your time when developing programs. You should have an entry for every time you program for this class. The columns should be self-explanatory except the phase column. The phase designations are listed on the right of the sheet. You may add your own phase codes and explanations if you think them necessary.

Before explaining the WBS, we should first explore how one goes about decomposing a problem. This process relies on your having gained a certain level of expertise: you should be able to determine what parts of the problem you understand how to accomplish and what you must work harder on. Here’s an example:

Write a program in your favorite language that reads in a series of floating point numbers, one to a line, and determines the minimum and the maximum. At the end of file, print out the minimum first and the maximum second.

Below is the transcript of a session that decomposes the above problem into components:

This is a full program, so there are three phases: input, process, and output. The process part is a while loop based on the “until end-of-file” clause. There’s no range specified, so I have to initialize the two variables, call them min and max, but I don’t have any particular value. OK, then read in the first one … Oops, note that there might not be a first one … init with the first one. Now you can have a loop with three statements: (1) check min changes, (2) check max changes, and (3) read next one.

Now the conditions on the while loop: need to have a variable—call it num_read—set by the input statement. Use num_read not zero to continue.

After the while loop, print out the two values. Oops, almost forgot the null case. Hm … there’s no spec on what to do there; print out nothing, I guess.

The approach given here is technically called talking a solution and this technique is very useful in decomposing a problem. You can use this technique yourself by either writing out your thoughts or by using a voice-activated tape recorder.

The WBS is just a form that aids the decomposition process. A logical way to begin is to decompose the problem and identify the components. It may be more natural to have a graphic representation of the components and their couplings (connections). Number each component and enter this as the “Task Number” in the form. Enter the component name and description. Fill in the “Inputs From” and the “Outputs To” columns based on the task number of the input and output components. Remember that at this level inputs may come from many different components and the outputs may go to many components. For all but the simplest projects, a graphical representation of this information is a must. We are now ready to estimate the effort.

Now estimate the number of “Lines” per component. The estimated “Hours” should be computed by either your measured rate or the 2.5 lines per hour rate that is the historical average. Now sum the “Hours” column to produce the cumulative hours, “Cum. Hours.”

We have now completed the planning portion. However, the WBS must be used during development to determine whether or not you can complete the project on time. In the next section we discuss earned value.

For illustration, suppose we have six modules A to F that must be delivered in two weeks. Suppose we estimate the number of time units (say hours) that this task should take as 100 and the amount of time needed for each as shown in Figure 11.3, called the baseline. In the baseline, the tasks are presented with the planned time estimates.

As the programming work is performed, value is “earned” on the same basis as it was planned, in hours. Planned value compared with earned value measures the volume of work planned versus the equivalent volume of work accomplished. Any difference is called a schedule variance. In contrast to what was planned, Figure 11.4 shows that work unit D was not completed and work unit F was never started, or 35 hours of the planned work was not accomplished. As a result, the schedule variance shows that 35 percent of the work planned for this period was not done.

Schedule variance really only makes sense if applied relative to completion when we also know exactly what must be done.

In this example, we’re really not in too bad shape. Four modules are complete and we estimated them correctly but we have badly miscalculated how long it takes to do the other two (D and F). The work still has to be done, but the next milestone looms large. The only thing you can do at this point is to consider merging the effort for D and F with what has to happen in the next milestone.

There are many lessons here:

The process of developing an earned value plan is exactly the process we need to design a program. The most important step is the WBS. Every milestone will start with developing a WBS, which includes everything that is required to successfully complete the project. The key to success is to account for everything! For example, the following classes of work need to be accounted for:

Note to the wise: Do not forget the effort for your other classes and extracurricular activities.

Because the design and implementation of each subprogram method and class structure take time, those times show on the WBS as well. If you are honest with yourself and honestly estimate your required effort, you will quickly see that you must be very disciplined. There is a sobering statistic: over the past 50 years, an accomplished programmer has been able to produce only 2.5 debugged statements per hour and this figure has been validated in several of the author’s classes.

As a procedure to develop each milestone:

The Personal Design and Software Process (PDSPP) documentation templates are provided as an Excel workbook. This section provides a key to that workbook based on the tab names in the Excel workbook. The original Excel workbook was retrieved online from the PSP Resources page at the University of Karlsruhe, Germany.^* There are many such sites on the Internet. The Time Utilization Log was developed for the Clemson University time schedules.

The Karlsruhe packet contains all the PSP forms mentioned in Humphrey’s book. The instructor should choose how much of the PSP discipline he or she is willing to impose. Note: Every form has a header that asks for what should be obvious information. For example, PSP0 asks for your name, due date, milestone, instructor, and language of implementation; this information should be obvious.