Over the past few weeks reading about unit testing, I’ve come across a few articles and blogs claiming that unit testing is not the most efficient way to ensure code accuracy and quality. I’ve been under the impression that unit testing is always a good idea, so I wanted to read about Robert Martin aka Uncle Bob’s input regarding this subject. He is a reputable computer programmer and the author of the well-received book “Clean Coder.” The following is a brief summary, along with my personal takeaways, from one of Bob’s blogs entitled First-Class Tests.
Bob is critiquing a TechBeacon blog where the author is condemning unit testing as bad practice. It seems to me that the TechBeacon blog author has claimed to “give up” unit testing for something that he calls “system testing.” But Bob insists that the TechBeacon author is not only running his tests in an inefficient way, but using improper terminology as well.
Bob gives a few definitions of different types of testing, some of which seem to contradict the views of the TechBeacon blog author. First, he explains that unit tests are written to ensure that the code is doing what the programmer intends it to do. Bob then gives his definition of system testing, stating it is not what the TechBeacon author believes it to be. System testing, according to Bob, are “plumbing” tests to ensure the internal workings of the entire integrated system is working as expected. This is in contradiction to the TechBeacon blog who claims he is running “system tests.” Bob claims what is actually being run here is “user-interface to database” functional tests, which are, in actuality, slow unit tests. I agree with Bob’s terminology here; it seems to coincide with what we’ve learned in class regarding unit testing thus far, some of which I’ve discussed in some of my previous blogs.
I believe the main point that Bob is trying to get across is that it seems many people tend to have the wrong idea of what unit testing is. If someone is writing unit tests that rely too heavily on outside resources, for example, it can slow down and hinder the whole purpose of testing software. It can make unit testing seem inefficient, when in actuality is the result of badly written unit tests. He stresses that programmers should not be trying to think of ways to circumvent unit testing. Instead, we should find ways to write unit tests as efficient, clean, and self-sufficent as possible.
I truly enjoyed reading Bob’s blog regarding his opinion on unit testing. I found it both amusing and informative. Bob leaves us with a final thought; we ought to treat our unit tests as “first-class citizens.” I will strive to do this throughout my professional career based on the advice he has given in his blog: I will avoid coupling unit tests with outside resources and make great efforts to always keep my unit test code “clean” and up-to-date.
From the blog CS@Worcester – Jason Knowles by Jason Knowles and used with permission of the author. All other rights reserved by the author.




