C++, "multiparadigm"??? - Re: NetBSD's i386 support, was Re: [cctalk] Old UNIX Systems

On Fri, 07 Jun 2013 19:50:58 -0400 Toby Thain <toby at telegraphics.com.au> wrote: [C++] Well. As allways with C++: It leaves the decision up to the programmer. This way the programmer can, well should, must, choose what way that is most efficient for the given situation. You can "plug in" a full blowen garbage collector to manage your memory. Or use smart pointers as a simple, primitive variation of gc. (... that works so well in many situations that a full blowen garbage collector is unnecessary many times.) You can pass the "lexical scope" by value, i.e. make a copy at the time the lambda is instantiated. (You may say that this defetas the purpose of lambdas. But actually it is the proper solution in many situations.) Or you can pass by reference. But in the later case you have to ensure that the reference is still valid when the lambda is used. That is the point about C++: If a language offers only one, single mechanism for a task, like garbage collection for memory management, it forces the programmer to this singe way. Even in situations where this one, single machanism is not apropriate. C++ deliberately gives you the choice of many ways. - And thus the responsibility to use your brains to find the way most efficient way for the speciffic, given task at hand. Thats why I dont like e.g. Java. It imposes garbage collection on me even where not appropriate or unwanted, making very usefull things like RAII de facto impossible. -- \end{Jochen} \ref{http://www.unixag-kl.fh-kl.de/~jkunz/}