RE: [Dev-C++] Newbie question Classes/Objects/Accessors????

[Smith, J. N. <jn...@ut...>]

-----Original Message-----
From: Smith, Jacob N. 
Sent: Tuesday, May 20, 2003 3:47 PM
To: 'Rus...@ao...'
Subject: RE: [Dev-C++] Newbie question Classes/Objects/Accessors????


> I understand the basics of Classes, but what determines which
> members/functions should be public/private? 

A simple (and effective) rule is to make everything private and only expose
those things which are needed for the client interface. Bruce Eckel has an
online book (for free) at www.bruceeckel.com for beginners. I'm sure others
could suggest more OOP-newbie books.

> of the same class. What is the logic behind this? I can
> understand what is 
> being done, but I don't understand the reasoning.

Sometimes a simple variable such as as "renderedTextLength( const Font&,
const std::string& )" has a simple meaning (what is the length of a string
using a font): with a fixed-width font the function is simple
::size()*width, for non-console fonts this must be calculated on the fly
using the font rules. By explicitly encapsulating all "value" calls (size,
width, height, et al) you (A) keep clients from accidently messing up your
class internals (B) provide a buffer so you can change implementation
details, e.g. consider the size function for an image. If the old image
class used a pointer to pixels then it's size is simply the length of the
array, if a new class uses a rectangular array then it's size is
width*height. If you'd exposed the height member to the clients you'd be
stuck with a specific implementation. Note that any sensible compiler will
be able to optimize the layer of indirection away so there is no overhead
for "accessor" methods.

-j.