At @forwardnetworks, we typically don't add such a test. Instead, in the body of the commit message, we include the reports from performance runs for a recent previous commit and the current commit.
@ploeh, do you use TDD to make performance improvements? That is, if you want to make some production code use less time, do you always write a test that fails without that optimization?
@nikosbaxevanis@ploeh@dharmaturtle@cmeeren Sorry for the delayed reply. We don't currently support Async<> for Task<>, but I think this is possible. I created an issue for this. https://t.co/KRpTPpwgn7
@ploeh I got it. Have the higher-order function that takes a function and returns a memoized function allocate (on the heap) the entire cache. Then it is not possible for the memoized function to have an out of memory exception (because of the memoization). Now this is a great example.
@KateJBaer, I think a coffee mug with covered up words that reappear when the mug gets hot would make a great medium for your erasure poetry. (Keep up the good work!) https://t.co/B8Na2aIWDs
@joshuamck@ploeh This can also be true for deterministic tests while doing TDD in a large code base because not all code is equally clear to all developers even when they made the test fail two minutes ago with their small code change.
@joshuamck@ploeh At my work, we phase the goal as making test failures actionable. I agree with Josh that long-running tests and non-deterministic tests are better when they have actionable failure messages.