While I agree with you on concurrency and partially on memory safety, there's a point I'd like to make.
Yes, compile-time computations is accomplished by templates. This is widely known as *template meta-programming*. However, *generic programming* is more like a programming paradigm. A way of thinking and building algorithms and concepts regardless of how the data structures they depend on is represented.
Generic programming does use some template meta-programming techniques. But in most cases it's done for type identification and not for the kind of computations you mention (remember the classical compile-time factorial). Although there's an overlap between these two areas, take the STL as generic programming in its essence and the Boost MPL on the other side.
Also, I'm not sure that templates were originally conceived as a replacement for macros. I think that even from the beginning there was more to it.
Slashdot Mirror
While I agree with you on concurrency and partially on memory safety, there's a point I'd like to make. Yes, compile-time computations is accomplished by templates. This is widely known as *template meta-programming*. However, *generic programming* is more like a programming paradigm. A way of thinking and building algorithms and concepts regardless of how the data structures they depend on is represented. Generic programming does use some template meta-programming techniques. But in most cases it's done for type identification and not for the kind of computations you mention (remember the classical compile-time factorial). Although there's an overlap between these two areas, take the STL as generic programming in its essence and the Boost MPL on the other side. Also, I'm not sure that templates were originally conceived as a replacement for macros. I think that even from the beginning there was more to it.