In a recent discussion about type-safe languages and the property that "it never crashes", I pointed out that these things run on fallible hardware and what's nicer than a system that almost never crashes is one that always recovers.
I fully agree with you. I’d only add that sometimes it’s hard to decide what to handle vs what to “let heal”, so IMO that’s when error tracking tools like Honeybadger come handy: when you’re notified of an unexpected / unhandled error, you can decide if it’s something you’d want to handle or if it’s fine to let it heal.
In a recent discussion about type-safe languages and the property that "it never crashes", I pointed out that these things run on fallible hardware and what's nicer than a system that almost never crashes is one that always recovers.
Maybe better said as "Let it recover"
I fully agree with you. I’d only add that sometimes it’s hard to decide what to handle vs what to “let heal”, so IMO that’s when error tracking tools like Honeybadger come handy: when you’re notified of an unexpected / unhandled error, you can decide if it’s something you’d want to handle or if it’s fine to let it heal.
why would i test if we just “let it crash”?
I hope this is a joke 😂