Almost every software engineering book:

Halting on an error is often best, as is raising the error after catching it, unless you're certain it should be subdued or aware, expect the issue, and have limited risk.

Every dev project:

Hold up, couldn't this raise an exception? That's bad!

Well, to be clear, I don't usually put people in danger. There's a lot these engineers could learn from software engineers: who are really the highest performance at engineering as a craft. One base rule: Above all, kill no one.

I think that software engineers put people in danger more often than they would like to believe; have a look at the UK Postal Service scandal for a great example of a seemingly innocuous bit of software absolutely destroying numerous lives.

Indeed. And yet immediately outmatched by the number of lives destroyed by motor vehicles (probably per day).

Neither the software nor the software engineers were the ones who destroyed people's lives. That's on managers of the software company, managers of the postal service, prosecutors and judges who all conspired to hide the truth and condemn innocent people. I believe software engineers actually testified that the software had bugs, and the prosecutors hid that particular testimony.