In none of the recipes in this chapter, we used exceptions to catch errors. While this is certainly possible, working on stream objects without exceptions is already very convenient. If we try to parse in 10 values, but this fails somewhere in the middle, the whole stream object sets itself into a fail state and stops further parsing. This way, we do not run into the danger of parsing variables from the wrong offset in the stream. We can just do the parsing in a conditional, such as if (cin >> foo >> bar >> ...). If this fails, we handle it. It does not appear very advantageous to embrace parsing in a try { ... } catch ... block.

In fact, the C++ I/O stream library already existed before there were exceptions in C++. Exception support was added later, which might be an explanation why they are not a first-class supported feature in the stream library.

In order to use exceptions in the stream library, we must configure each stream object individually to throw an exception, whenever it sets itself into a fail state. Unfortunately, the error explanations in the exception objects, which we can then catch later, are not thoroughly standardized. This leads to not really helpful error messages, as we will see in this section. If we really want to use exceptions with stream objects, we can additionally poll the C library for filesystem error states to get some additional information.

In this section, we are going to write a program that can fail in different ways, handle those with exceptions, and see how to squeeze more information out of those afterward.