--- Luke Dickens <lwd03@xxxxxxxxxxxx> wrote: > Thank you Ted. > I was going to look at smart pointers when I did a > code clean-up. I have > the boost libraries, but don't want to introduce too > much new code all > at once. Are they fairly easy to incorporate into > existing code, or is > this too pervasive a coding paradigm requiring an > entire rework. > In my code, they are pervasive as I use them from the start, as planned during the design phase of a project. That said, how easy or difficult it is to refactor your code to use this idiom depends on how much code there is, and how you use the objects. But, I have seen it take just a few hours on a project having 500,000 lines of code. The kinds of issues, if the existing code takes objects of type T, range from overloading your functions that work on them to take smart pointers to type T, that the correct operator '->' vs '.' is used to invoke member functions, the right function objects are used, &c. Everything you do with these objects would need to be recast as working with smart pointers instead. But doing it all is relatively easy if you have a decent development IDE and you are organized. You DO have design documents that specify what you do to what, and where the relevant code is, right? If so, that makes it trivial to construct a to do list that can serve to check off all the changes you'd need to make to use smart pointers instead of the objects of type T themselves. That kind of organization makes most things go much faster. HTH Ted
