Subj : Re: "File Scope" objects
To   : comp.programming,comp.software-eng
From : Phlip
Date : Thu Aug 18 2005 03:23 pm

Gerry Quinn says...
>
>> For C++, many classes should live only inside a .cpp file. (Put another 
>> way,
>> many C++ newbies think that all class definitions only belong in header
>> files, which is wrong.)
>
> You want to define class B in the implementation file of class A, i.e.
> "a.cpp".
>
> But suppose class B is written as normal, i.e. "b.h" and "b.cpp", but
> "b.h" is included only in "a.cpp".
>
> You won't be looking at "b.h" anyway except by way of finding an
> include in "a.cpp".  So why worry about it?

Because #include "b.h" is a compilable comment tempting your colleagues 
"b.h" is designed to be reused.

It is not.

I am aware there are other ways to comment the situation, and there are 
other technical reasons to #include "b.h".

> Doing thing the normal way

There's a difference between a normalized way and "the way we have always 
done it".

> gives effectively the same result, and stops classes contaminating each
> others files.
>
> Only if you are sure B will be useless without A do you gain anything -
> and even then, you have a longer, more unwieldy file, in which
> functions of two classes have to be kept in order.  Or you use
> something like MSVC that obscures the details of file structure by
> creating hyperlinks to functions, in which case there is also no
> benefit.

In some situations long .cpp files speed up compiling and linking. There are 
so many physical issues involved that we cannot promote the idea "put the 
private class B entirely inside A's implementation file" to the level of a 
"guideline". It's still a helpful suggestion, and a way to debunk the newbie 
that all classes belong in .h files.

-- 
  Phlip
  http://www.greencheese.org/ZeekLand  <-- NOT a blog!!!

.