There is an episode of Family Guy where Peter Griffin gets a segment on the Quahog 5 News called “What Really Grinds My Gears.“ During the segment, he opines on random things that bother him in his own hilarious and offensive way.
When it comes to software projects and technology, I have my own list of pet peeves that continues to grow. While not quite as hilarious though potentially just as offensive to some, here are a few.
JAVA and SCRUM
Stop yelling at me! Java is the most popular programming language in the world. Scrum is the most popular agile framework in the world. So why do so many people, including software engineers who should know better, still fail to understand that neither is an acronym? If I ran Stack Overflow, I would impose a lifetime ban on anyone who uses all caps for Java and Scrum.
Speaking of Java, enough already with calling the Java Platform, Enterprise Edition “J2EE.” J2EE hasn’t been a thing since May 2006! That’s the same year the first Blu-Ray movies were sold. Some people even still liked Barry Bonds as he broke Babe Ruth’s home run record. There is no excuse for anyone—especially software engineers—to use J2EE anymore.
Of course there are still Java Enterprise API’s, but the platform has been officially known as Java EE for eight years. If you must, you can even use JEE. Either way, just stop with J2EE. If you can come to terms with P. Diddy and Snoop Lion, you can come to terms with Java EE.
But software engineers just write code. That’s it. So many of them on Stack Overflow and elsewhere begin their questions with “I wrote a code and…” This offense would also earn a lifetime ban from me. If you really want to write “a code,” get into cryptography. Programming is just code.
Referring to HTML and CSS “code”
Speaking of code…
As described in the Vidya tutorial Web Design 101, HTML and CSS are two pillars of the web. They are powerful technologies that have changed our daily lives. Neither, particularly CSS, is easy to master.
But HTML and CSS are not code.
”I believe in the right tool for the job”
“I believe we should keep things simple”
If a politician says, “I believe in a robust economy and a stout national defense,” he or she has provided you with an empty, meaningless platitude no one can possibly argue with, which is probably the point. But as you know, it’s the details that matter.
”I believe in the right tool for the job” and “I believe we should keep things simple” are the equivalent in software projects.
No one actively seeks out the wrong tool for the job. No one tries to use JUnit to run a MapReduce job. Still, people can reasonably disagree over the merits of particular technologies for solving particular problems. If you are building a non-trivial web application, should you use Rails, Django, or Grails? All three are preeminent tools, but which is the “right tool for the job”? How do you know?
Similarly, no one wants to add complexity to a project, but people can reasonably disagree over what constitutes complexity. On a Java project talking to a relational database, is it simpler to access the database with pure JDBC but manage transactions and write all the tedious object-relational mapping (ORM) yourself? Or to add Hibernate as a dependency to your build file, learn it, and configure it but let it handle the responsibility for transaction management and ORM for you?
Not so obvious, is it? This is why anyone who uses these platitudes on a software project must be compelled to explain a specific course of action and the rationale for it.
Managers who want to see the metrics for unit tests
As I wrote in my first column for Government Computing News, unit tests are by developers for developers. Unit tests are essential for helping developers confirm their code works as expected. Unit tests also help developers refine the design of their code. There is an inflection point though where unit tests start to bring diminishing returns.
A common theme of the TV classic The Wire is that senior management in various Baltimore institutions demand their people achieve useless performance metrics that ultimately prove counterproductive to their missions. Just like the police officers, teachers, and others in The Wire, software engineers are motivated to “juke the stats“ to appease managers who demand endless unit testing because of cool things they heard at conferences. So developers write prolific tests that don’t actually test anything but produce a high number of passing results that make these micromanagers happy. The insidious result is that the applications decay more and more while they appear healthier and healthier. Everyone, including the managers, will have lots of ‘splainin to do at release time.
Smarter managers will instead trust their engineers to do what they know best and look instead at acceptance test results and code coverage reports. Just stay away from anything related to unit tests.
So there you have it. All these things really grind my gears. And there are more…