alt Hacker NewsYea but instead of log Critical/Fatal and go on, I would just panic() the program. To the other definitions I agree - everything else is recoverable, because the program still runs.
Warning to me is an error that has very little business logic side effects/impact as opposed to an Error, but still requires attention.
I write a lot of backend web code that often talks to external services. So for example the user wants to add a shipping address to their profile but the address verification API responds with a 500. That is an expected error: sometimes it can happen. I want to log it but I do not want a trace back or anything like that.
On the other hand it could be that the API had changed slightly. Say they for some reason decided to rename the input parameter postcode to postal_code and I didn’t change my code to fix this. This is 100% a programming error that would be classified as critical but I would not want to panic() the entire server process over it. I just want an alert that hey there is a programming error, go fix it.
But what could also happen is that when I try to construct a request for the external API and the OS is out of memory. Then I want to just crash the process and rely on automatic process restarts to bring it back up. BTW logging an error after malloc() returns NULL needs to be done carefully since you cannot allocate more memory for things like a new log string.