What Do I Tell My Manager?

How to tell a manager about refactoring is one of the most common questions I've been asked. If the manager is technically savvy, introducing the subject may not be that hard. If the manager is genuinely quality oriented, then the thing to stress is the quality aspects. Here using refactoring in the review process is a good way to work things.

Tons of studies show that technical reviews are an important way to reduce bugs and thus speed up development. Take a look at any book on reviews, inspections, or the software development process for the latest citations. These should convince most managers of the value of reviews. It is then a short step to introduce refactoring as a way of getting review comments into the code.

Of course, many people say they are driven by quality but are more driven by schedule. In these cases I give my more controversial advice: Don't tell!

Subversive? I don't think so. Software developers are professionals. Our job is to build effective software as rapidly as we can. My experience is that refactoring is a big aid to building software quickly. If I need to add a new function and the design does not suit the change, I find it's quicker to refactor first and then add the function. If I need to fix a bug, I need to understand how the software works—and I find refactoring is the fastest way to do this. A schedule-driven manager wants me to do things the fastest way I can; how I do it is my business. The fastest way is to refactor; therefore I refactor.

Indirection and Refactoring

Kent Beck

Computer Science is the discipline that believes all problems can be solved with one more layer of indirection. —Dennis DeBruler

Given software engineers'infatuation with indirection, it may not surprise you to learn that most refactoring introduces more indirection into a program. Refactoring tends to break big objects into several smaller ones and big methods into several smaller ones.

Indirection is a two-edged sword, however. Every time you break one thing into two pieces, you have more things to manage. It also can make a program harder to read as an object delegates to an object delegating to an object. So you'd like to minimize indirection.

Not so fast, buddy. Indirection can pay for itself. Here are some of the ways.

  • To enable sharing of logic.

    For example, a submethod invoked in two different places or a method in a superclass shared by all subclasses.

  • To explain intention and implementation separately.

    Choosing the name of each class and the name of each method gives you an opportunity to explain what you intend. The internals of the class or method explain how the intention is realized. If the internals also are written in terms of intention in yet smaller pieces, you can write code that communicates most of the important information about its own structure.

  • To isolate change.

    I use an object in two different places. I want to change the behavior in one of the two cases. If I change the object, I risk changing both. So I first make a subclass and refer to it in the case that is changing. Now I can modify the subclass without risking an inadvertent change to the other case.

  • To encode conditional logic.

    Objects have a fabulous mechanism, polymorphic messages, to flexibly but clearly express conditional logic. By changing explicit conditionals to messages, you can often reduce duplication, add clarity, and increase flexibility all at the same time.

Here is the refactoring game: Maintaining the current behavior of the system, how can you make your system more valuable, either by increasing its quality or by reducing its cost?

The most common variant of the game is to look at your program. Identify a place where it is missing one or more of the benefits of indirection. Put in that indirection without changing the existing behavior. Now you have a more valuable program because it has more qualities that we will appreciate tomorrow.

Contrast this with careful upfront design. Speculative design is an attempt to put all the good qualities into the system before any code is written. Then the code can just be hung on the sturdy skeleton. The problem with this process is that it is too easy to guess wrong. With refactoring, you are never in danger of being completely wrong. The program always behaves at the end as it did at the beginning. In addition, you have the opportunity to add valuable qualities to the code.

There is a second, rarer refactoring game. Identify indirection that isn't paying for itself and take it out. Often this takes the form of intermediate methods that used to serve a purpose but no longer do. Or it could be a component that you expected to be shared or polymorphic but turned out to be used in only one place. When you find parasitic indirection, take it out. Again, you will have a more valuable program, not because there is more of one of the four qualities listed earlier but because it costs less indirection to get the same amount from the qualities.