Maybe even more notable, at sxsw I went to a panel on scaling webapps and was given this exact line of reasoning by some obviously competent dev leads from successful internet companies. "We're a startup. We're resource-constrained. We don't have time to write tests".
I don't want to ridicule these people - you have your set of experiences and you can't have absorbed everything the development world has to offer.
But we know that we write tests precisely because we're time-constrained. I think this should get more emphasis than it does: a well-crafted, fast-running unit test suite is an incredibly powerful productivity tool. If your time horizons are > 1 month, your team will spend a punishing amount of time reorienting the codebase to accommodate changes and new features if correctness must be proven at human-using-a-browser speed. For a modern startup it could very well mean death.
Rob Myers
· 8 months ago
Perfect analogy!
I just gave a talk last night on the business value of TDD and particularly refactoring, and I find that--in order to get real buy-in--I've had to finish this particular talk with three actual first-person stories about how the team I was on turned a sudden, surprising, and "architecturally challenging" story into over $100,000/year savings or revenue in a matter of days. I know for a fact that all three of those would have ended in disappointment if we hadn't been completely covered with fast, automated regression tests, or hadn't been refactoring with a very low tolerance for subtle code-smells. (That was one point of the talk I aimed at developers: On a TDD team, you're not refactoring messy code, you're cleaning CLEAN code. Like a sushi kitchen, you never leave it even a little messy.)
All Comments
I don't want to ridicule these people - you have your set of experiences and you can't have absorbed everything the development world has to offer.
But we know that we write tests precisely because we're time-constrained. I think this should get more emphasis than it does: a well-crafted, fast-running unit test suite is an incredibly powerful productivity tool. If your time horizons are > 1 month, your team will spend a punishing amount of time reorienting the codebase to accommodate changes and new features if correctness must be proven at human-using-a-browser speed. For a modern startup it could very well mean death.
I just gave a talk last night on the business value of TDD and particularly refactoring, and I find that--in order to get real buy-in--I've had to finish this particular talk with three actual first-person stories about how the team I was on turned a sudden, surprising, and "architecturally challenging" story into over $100,000/year savings or revenue in a matter of days. I know for a fact that all three of those would have ended in disappointment if we hadn't been completely covered with fast, automated regression tests, or hadn't been refactoring with a very low tolerance for subtle code-smells. (That was one point of the talk I aimed at developers: On a TDD team, you're not refactoring messy code, you're cleaning CLEAN code. Like a sushi kitchen, you never leave it even a little messy.)