1) Why cant I just throw an boost::exception object?
throw boost::exception() << boost::error_info<tag_errno>(100) << boost::error_info<tag_file_name>("hello");
There are two reasons: 1) This design encourages throwing different types for different failures, which is a good programming practice 2) It protects against the following subtle error: try { foo(); } catch( boost::exception & x ) { x << boost::error_info<tag_file_name>("hello"); throw x; //should have been just throw; } Instead of re-throwing the original exception object by a throw expression with no operand, we accidentally throw a new exception object, which erases the original exception type (in the current design of the library, this situation results in compile error.)
2) It might be nice to support "std::cerr << e", instead of requiring the use of e.what().
std::cerr << e;
Yes, though I'm not big fan of operator overloading. OTOH, this can be useful in templates, so I think it's a good idea.
3) Support for linking multiple error tags in a single expression might be useful and help in standarizing errors throughout an application.
throw boost:::exception() << boost::error_info<tag_errorno,tag_filename,tag_size,tag_width>( (1,"output.txt",100,200);
Could you clarify what you mean? Thanks for your feedback! Emil Dotchevski