Digging a hole
There’s a tale of how to dig a hole.
If you watch a professional dig a large hole they don’t just dig straight down. They build themselves a little set of steps. First one step, then they’ll stand on that one to dig the next, and so on until they reach a depth they’re aiming for, and they’ll dig away the steps.
Don’t be afraid to dig away your steps (intermediate tests) once they’ve got you where you’re going. Developers are often reluctant to delete code, even if it was code they just wrote themselves.
- If you’re afraid of losing something worthwhile, commit it to version control. The delete it.
- If it was code written by someone else:
- is it obvious this test is now not worthwhile? Delete it.
- is the original developer still available to you? - chat to them about the purpose of the tests. Is that purpose still necessary. Then delete it.
- are they not here? What are you afraid of? Delete it.
I am NOT advocating deleting tests willy-nilly. I’m talking about trivial, starter tests that have outgrown their usefulness. Stuff like:
- Constructor tests (by the time you’ve got a test harness you’ll have made test constructors that define this behaviour)
- Exploratory tests
Once they’ve been committed, let them go.
People will be able to find them in the future with
git log -S"<magic line of code that I had committed 6 months ago>"