Re: New release of the Dynace OO extension to C

BGB / cr88192 wrote:


  I'm honestly a bit puzzled by that opinion.

  First you learn that RTTI is not mandatory in C++, which makes
instances of the class spacewise optimal (very similar to a struct
containing the same member variables), but then you still say that you
prefer not to use a class in these cases. Why not?

  The advantage of a C++ class (with no RTTI) over a C struct is that
you can have a public and a private interface, which increases
abstraction and modularity. You can ever *inherit* from such a class
(and still have no RTTI overhead), which gives some design advantages.

  You could achieve almost the same result by using a plain C struct and
a bunch of functions taking instances of it as parameter (as a
substitute for member functions), but in that case you are lessening the
modularity and abstraction of that construct. (C structs also lack other
beneficial properties of C++ classes/structs, such as constructors and
destructors, which make them a lot easier to use eg. in arrays, not to
talk about safety if the struct has eg. pointers to dynamically
allocated memory or such.)

  A C++ class, even without RTTI, is a very powerful tool. You could
have, for example, a string class which automatically allocates and
deallocates itself (when it goes out of scope), automatically copies
itself when passed by value (using any of a number of techniques, eg.
the copy-on-write mechanism), checks for access boundaries, etc. All
this without any space overhead from RTTI. Thus the space taken by one
of these string objects can be the same as a char* (which is what you
would usually use in C).

  I honestly don't understand why so many C programmers are so
prejudiced against C++. C++ is a wonderful expansion to C.