Any good arguments for not writing unit tests?

Question

question from:https://stackoverflow.com/questions/2933391/any-good-arguments-for-not-writing-unit-tests

Answer 1

In the places I have previously worked, unit testing is usually used as a reason to run with a smaller testing department; the logic is "we have UNIT TESTS!! Our code can't possibly fail!! Because we have unit tests, we don't need real testers!!"

Of course that logic is flawed. I have seen many cases where you cannot trust the tests. I have also seen many cases where the tests become out of date due to tight time schedules - when you have a week to do a big job, most developers would spend the week doing the real code and shipping the product, rather than refactoring the unit tests for that first week, then pleading for at least another week to do the real code, and then spending a final week bringing the unit tests up to date with what they actually wrote.

I have also seen cases where the business logic involved in the unit test was more monstrous and hard to understand than the logic buried in the application. When these tests fail, you have to spend twice as long trying to work out the problem - is the test flawed, or the real code?

Now the zealots are not going to like this bit: the place where I work has largely escaped using unit tests because the developers are of a high enough calibre that it is hard to justify the time and resource to write unit tests (not to mention we use REAL testers). For real. Writing unit tests would have only given us minimal value, and the return on investment is just not there. Sure it would give people a nice warm fuzzy feeling - "I can sleep at night because my code is PROTECTED by unit tests and consequently the universe is at a nice equilibrium", but the reality is we are in the business of writing software, not giving managers warm fuzzy feelings.

Sure, there absolutely are good reasons for having unit tests. The trouble with unit testing and TDD is:

• Too many people bet the family farm on it.
• Too many people use it as a religion rather than just a tool or another methodology.
• Too many people have tried to make money out of it, which has skewed how it should be used.

In reality, it should be used as one of the tools or methodologies that you use on a day to day basis, and it should never become the single methodology.