[Dev-C++] Re: Design decision in C++

[Kevin B. <kev...@ma...>]

The main advantage to the static idea, of course, is that there will be no 
"this" pointer passed to all your calls, which will make them slightly more 
efficient.
The disadvantages are that the class will be very difficult to generalize 
and reuse, and that the syntax will seem "weird" and probably a bit 
difficult to understand for anybody else who ever needs to maintain this class.

Personally, I would disregard the "efficiency advantage", since it will 
likely be utterly insignificant. You should consider an "optimization" like 
this only when you have clear evidence from performance profiling of a 
working program that it will make some difference. The key is objective 
evidence; developers are notoriously bad at guessing where their 
performance bottlenecks lie. Resist the temptation to over-optimize before 
you know what the bottlenecks are, and especially before you even have a 
working design. The rule is, first make it correct, then make it fast.

If you're going to make all the class members static, you might as well 
just forget the class and define plain old variables and functions in a 
namespace. That would be syntactically almost identical, and conceptually 
cleaner. This all-static idea is *not* the same as a singleton pattern. A 
singleton class in general has no more than one instance at a time. With an 
all-static class, by contrast, there's no point having any instance at all. 
There's nothing for the "this" pointer to point to, so why bother? Ergo, if 
you're going to go this route, you might as well just use a namespace 
instead and avoid confusing future maintainers (and maybe even yourself) 
with a pointless class declaration.

--- Kevin