We did some complex input retrieval in this section. The first noticeable thing is that we always did the retrieval and error checking at the same time.

The result of the expression cin >> x is again a reference to cin. This way, we can write cin >> x >> y >> z >> .... At the same time, it is possible to convert it into a Boolean value by using it in a Boolean context such as if conditions. The Boolean value tells us if the last read was successful. That is why we were able to write if (cin >> x >> y) {...}.

If we, for example, try to read an integer, but the input contains "foobar" as the next token, then parsing this into the integer is not possible and the stream object enters a fail state. This is only critical for the parsing attempt but not for the whole program. It is okay to reset it and then to try anything else. In our recipe program, we tried to read a list of names after a potentially failing attempt to read two numbers. In the case of a failing attempt to read those numbers in, we used cin.clear() to put cin back into a working state. But then, its internal cursor was still on what we typed instead of numbers. In order to drop this old input and clear the pipe for the names input, we used the very long expression, cin.ignore(std::numeric_limits<std::streamsize>::max(), 'n');. This is necessary to clear whatever is in the buffer at this point, because we want to start with a really fresh buffer when we ask the user for a list of names.

The following loop might look strange at first, too:

for (string s; getline(cin >> ws, s, ',');) { ... }

In the conditional part of the for loop, we use getline. The getline function accepts an input stream object, a string reference as an output parameter, and a delimiter character. By default, the delimiter character is the newline symbol. Here, we defined it to be the comma (,) character, so all the names in a list, such as "john, carl, frank", are read individually.

So far, so good. But what does it mean to provide the cin >> ws function as a stream object? This makes cin first flush all the whitespace, which lead before the next non-whitespace character and after the last comma. Looking back at the "john, carl, frank" example, we would get the substrings "john", " carl", and " frank" without using ws. Notice the unnecessary leading space characters for carl and frank? These effectively vanish because of our ws pretreatment of the input stream.