Exceptions in C : Should We Specify Them in Function Prototypes?

In C , exception specifiers allow functions to declare whether they may throw exceptions. However, their use has been questioned due to concerns about their effectiveness and consequences.

Reasons Against Using Exception Specifiers:

• Weak Enforcement: The compiler doesn't strictly enforce exception specifiers, so violating them may not result in an error. This undermines their reliability.
• Program Termination: Violating an exception specifier typically triggers program termination, which is an unfavorable outcome in most cases.
• Standard Non-Compliance: Some compilers, such as VS.Net, treat throw(X) as throw(...), which violates the standard and can lead to unreliable behavior.

Reasons to Avoid Exception Specifiers:

• Difficult to Implement in Templates: Exception specifiers hinder template code because it's challenging to anticipate all possible exceptions that might arise.
• Prevent Extensibility: Specifying specific exceptions can restrict the future evolution of a function's behavior due to its potential impact on existing code.
• Uncertainty in Legacy Code: Interfacing with external libraries that may throw unanticipated exceptions can make it difficult to properly handle errors and ensure the stability of the program.

Alternatives to Exception Specifiers:

Instead of using exception specifiers, it's generally advisable to:

• Return error codes or objects indicating status or specific conditions.
• Catch exceptions and handle them in a controlled manner.
• Rely on try-catch blocks to trap and handle exceptional cases explicitly.

In conclusion, while exception specifiers provide a mechanism to communicate the potential for exceptions, their limitations and potential negative consequences make them less advisable for general use. By adopting alternative strategies for error handling, developers can enhance the robustness and maintainability of their code.