An exception is an instance of a special type of class called Exception, or one of its sub-classes. They indicate when something goes wrong in our programs. Most importantly, if our program raises one of these exceptions, they will crash, unless we handle them.
What does it mean to ‘handle’ an error?
Well, in programming, sometimes our applications will throw errors that are out of our control. In these situations, it’s our responsibility to handle the potential error that may surface, and the more precise we can be in doing so, the better.
Precise…?
Yes. There are all kinds of different exceptions that can be raised, but only a subset of these that our programs can actually be responsible for. All of those in which our program can potentially throw are within the StandardError subclass of the Exception class. There are subclasses within the StandardError subclass as well. The goal is to always rescue the most specific type of error that we expect in which also encapsulates all possible errors. For example, a sub-class of StandardError is IndexError which has two sub-classes- KeyError and StopIteration. Therefore, if we were rescuing for a potential error that could be either a KeyError or a StopIteration error, then we would rescue for IndexError. However, if the only potential error was a KeyError, then we would just rescue for KeyError.
The bottom-line
We should never rescue for the generic Exception class, because within the Exception class, there are errors that are impossible for our programs to raise. All of the errors that our programs could potentially be responsible for raising are are sub-classes of the StandardError class, and therefore should be the most generic type of error that we ever rescue for.