When people start writing some new code they often feel so enthusiastic that they try to make it reusable right from the beginning. Or some do it out of some pretended sophistication. The end result is usually layers of unnecessary abstractions that are actually not used anywhere else. In the end such code takes more effort to write and more effort to maintain afterwards.
The reality is that you cannot make something reusable if it is used in a single place/scenario. To make it reusable you do need at least two real use cases, so you can abstract the common traits. The more use cases you have, the better you can refine the abstraction.
You may think you know what new use cases will come, but in reality these either don't come at all or are quite different from what you have expected.
So don't do this, unless you can tell the future.
I thought premature generalization would be a good term for such situations. Then after searching the net it turned out this term is well established already - Premature Generalization Is Evil. It is even listed as #1 in Seven deadly sins of programming.
No comments:
Post a Comment