That’s a really interesting observation actually and it always plays out in my mind the same way it has in this example – I don’t remove the duplication until I get a nagging feeling that I should be doing it right at the end.

And then there’s 2 extra lines at the top of the test which don’t really tell us anything other than what we can already tell from glancing at the test – thanks for pointing that out though!

I wonder when it’s important to actually extract local variables though – my current thinking (for non strings) is that if I care about that bit of data for the test I want to try and extract the value into a more descriptive name but if not then I’m happy to leave it inline.

I’d be interested in knowing your thoughts.

Mark

I don’t know why you take the last step. You’re not alone — for some reason everyone feels the need to pull out local variables for duplicated string constants. But I think it harms rather than improves the readability of the test, and that should be the test, rather than blindly trying to minimize duplication.

I think that deep down inside you feel the same way, which is why you say “if we want to remove duplication” rather than just going ahead and doing it. Have the courage of your convictions — shake off the shackles of reflexive deduplication.