I try to write effective tests that really help me out without being too constraining or tedious.<p>When I go back to fix something or refactor old code, I know what an anxious experience it is to do that without test coverage, so having a basic test suite makes me happy.
MVPs? Absolutely. Just because it's the first iteration doesn't mean it's not important to test that everything in that first iteration works.<p>Hobby projects? Less so, particularly if it's just me writing software. However, if I'm writing a particularly hairy piece of logic I'll absolutely write some tests for it (for example, I wrote a JS calendar plugin just to see if I could do better than the many existing ones, and I did all of the date logic myself using JS's terrible Date library).
Yes. It's a bit hard to do so for games, but doing so makes it SOOOOOOO much easier to port if you have to. After all a test is basically a port to a system that has a different graphical, windowing & input system (none, none, & simulated).
My friend shared tons of details in how he tested his future wife for possibility of long lasting relationship.<p>They got divorced 1 yr after marriage.<p>So much for writing a test cases. Follow your intuition.
If I intend to share or build upon it, yes.<p>If I'm exploring an idea that I know is throwaway and won't be foundational for anything else, no.