Error handling in C++, constructors vs. regular methods

Question

I have a cheesesales.txt CSV file with all of my recent cheese sales. I want to create a class CheeseSales that can do things like these:

CheeseSales sales("cheesesales.txt"); //has no default constructor
cout << sales.totalSales() << endl;
sales.outputPieChart("piechart.pdf");

The above code assumes that no failures will happen. In reality, failures will take place. In this case, two kinds of failures could occur:

• Failure in the constructor: The file may not exist, may not have read-permissions, contain invalid/unparsable data, etc.
• Failure in the regular method: The file may already exist, there may not be write access, too little sales data available to create a pie chart, etc.

My question is simply: How would you design this code to handle failures?

One idea: Return a bool from the regular method indicating failure. Not sure how to deal with the constructor.

How would seasoned C++ coders do these kinds of things?

Answer 1

In C++ exceptions is the way to report errors. BTW exception in initialization list CAN be handled.

A function-try-block associates a handler seq with the ctor-initializer, if present, and the function-body. An exception thrown during the execution of the initializer expressions in the ctor-initializer or during the execution of the function-body transfers control to a handler in a function-try-block in the same way as an exception thrown during the execution of a try-block transfers control to other handlers.

Good code usually should use minimum of try/catch blocks on most upper (thread) level. Ideally only one. This way, knowing that "everything throws", you have not think too much about errors and your normal scenario code flow looks clean.

Answer 2

A constructor is used for initializing object's internal state. Do not use it for making heavy operations like reading and processing files.

Use a method instead which will read the file and throw an exception (or return a boolean for success) incase an error occured. Catch this exception in your main flow and handle it as you see fit.

EDIT : If this is the whole class purpose than maybe ChessSales should only contain the data and you should use a factory class (or maybe static utility class) which has a method for reading a CSV file and returning a ChessSales object containing the relevant data read from the CSV file. This way you separate your data from the business logic (in this case reading and parsing the CSV file)