Refactoring Boundary

1 September 2004

There was some recent discussion on the refactoring mailing list about what is or isn't a refactoring. As with these discussions, there's always a danger of debating how many angels fit on a pin, but thinking about the boundaries does have some useful purpose.

The DefinitionOfRefactoring in my book was intentionally informal. The informality rests on a couple of phrases that are distinctly open to interpretation:

  • without changing its observable behavior: which begs the question what is observable behavior? Essentially it means that the software still does what it did before - but there's lots of ways you could interpret that.
  • to make it easier to understand and cheaper to modify: this gets at the purpose of refactoring. There are many changes we can make to our programs, but in my view refactoring is all about making it easier to understand and change. The same changes made with a different purpose aren't refactoring as I see it.

The essence of refactoring is that of the sequence of small behavior-preserving changes. Despite the fact that refactoring isn't something that can be formally defined, it's still a pretty precise terms - and I do want to avoid RefactoringMalapropism. But I think it's worth thinking about some of these cases, which I'm putting in different bliki entries: IsChangingInterfacesRefactoring, IsFixingAnUnknownBugRefactoring, IsOptimizationRefactoring, and IsDeclarationOrderingRefactoring.