The library provides an exception class that can transport arbitrary values, thus freeing the user from designing custom exception classes.
I should clarify that the purpose of the library is not to free users from writing custom exception classes: after all, exceptions are caught by type -- not by value -- so to be able to tell one failure from another, we need to throw different types. The idea behind the proposed exception library is to move the burden of storing error codes, file names and other data relevant to a given failure, from the individual exception classes, to a single base class. This way the exception class hierarchy can be implemented entirely in terms of empty classes while still being able to transport to the catch site whatever information is necessary to handle a particular exception. Emil Dotchevski