I tend to do that mostly for concurrent programs, as I want to see what happens without blocking the entire system like a debugger would.

For medium/large programs I'll often prefer a debugger because I can easily and quickly try to diagnose the problem without having to wait for a new build to complete with the added traces.