How should you design custom exception types in OO languages?

How should you design custom exception types in OO languages?

• A. Do not create new exception types; use a generic error.
• B. Only declare exceptions in the API; never implement new ones.
• C. Use a non-specific base exception for all errors.
• D. Create specific exception classes for distinct error conditions, derive from a common base, and provide meaningful messages to aid debugging.

Answer: (D)

Model exceptions as a small typed hierarchy where each distinct error condition has its own class. All of these classes should derive from a common base exception so they’re part of the same family, which makes it possible to handle them together when needed or catch a specific one for targeted recovery. Each exception class should convey a meaningful message (and can carry extra context like codes or relevant data) to help developers understand what went wrong. This design gives precise, actionable information: you can catch specific failures to respond appropriately, or catch the base type to handle a whole class of related errors in a uniform way. Using a single generic error or a non-specific base for all errors hides the exact problem and makes debugging harder, and relying only on API declarations without implementing distinct exception types limits extensibility and clarity.

Model exceptions as a small typed hierarchy where each distinct error condition has its own class. All of these classes should derive from a common base exception so they’re part of the same family, which makes it possible to handle them together when needed or catch a specific one for targeted recovery. Each exception class should convey a meaningful message (and can carry extra context like codes or relevant data) to help developers understand what went wrong. This design gives precise, actionable information: you can catch specific failures to respond appropriately, or catch the base type to handle a whole class of related errors in a uniform way. Using a single generic error or a non-specific base for all errors hides the exact problem and makes debugging harder, and relying only on API declarations without implementing distinct exception types limits extensibility and clarity.