I think every software developer views planetary landings with extreme dread.

We've (developers) all have extensively tested stuff we were proud of and expected it to work. And 100% of the time, something will be f%$%ed that needs debugged. Imagine relying on 10,000 individual systems that are, in fact, no better than alpha tested until it finally gets it's ultra-billion dollar ass to another planet, with telemetry delays of over 15 minutes (speed of light over, and back). And by alpha tested, I mean there is never an actual, complete, working test of all components together until it's there. Just "simulated". Here, something starts to go wrong you could pull the plug. There, dust is already settling from it's impact crater by the time the information gets to you that something was wrong at the beginning of its decent.

It is honestly mind blowing that there isn't more cases of "failed to convert between metric and imperial", or more likely, "some dingleberry forgot a f$#$ing semicolon" and "his dip$h!t friend didn't close a rarely-used loop" in the space program, considering how much code must go into these things now.