How LoD is useful.
in [OOP]
|
Prev: How LoD is useful. (was: DbC, OOD and Stroustrup's Rule...)
Next: Pattern for user rights on form's objects
From: Dmitry A. Kazakov on 7 Aug 2007 04:00 On Mon, 06 Aug 2007 18:56:29 GMT, Daniel T. wrote: > John Carter <joNhOn.ScPaArMteErV(a)EtRait.co.nz> wrote: > >> ie. Ways in which bad designs allow us to subtly bypass the invariant >> protection. > > One of my favorite examples: > > struct Point { > int x; > int y; > }; > > class Rectangle { > public: > Point& getTopLeft(); > Point& getBotRight(); > > // invariant: > // getTopLeft().x < getBotRight().x && > // getTopLeft().y < getBotRight().y ); > }; > > void fn1( Point& p ) { > p.x = 455; > } > > void fn2( Rectangle& r ) { > fn1( r.getTopLeft() ); > } > > Was the Rectangle object's invariant broken? Maybe, maybe-not... The design is broken. In this design, there is a getter/setter pair represented by getTopLeft(). The setter of this pair violates the invariant. > There > is nothing inherently wrong with the two functions, however in > combination, they are forcing a change in the state space of the > rectangle without proper permission. Yes and a proper design implies that no any combination can break the invariant, as DbC requires. The program above is simply incorrect from DbC stand point. Though it can be correct in a wider sense, without DbC. > However, most languages don't have the concept of a "constant > reference". This is irrelevant. The way getter/setter are implemented is up to the compiler. This cannot influence program correctness. In your example, the compiler could well do copy-out/copy-in for getTopLeft(), which by the way will be by wide margin more effective on most modern processors, than using references. It is a big optimization problem in C++, that references are published. One should never ever do that in presence of CPUs with multi-level cache. There are lots of object for which you just cannot get the address of. Or consider a massively networking application with Rectangle on a different host. What is Point&? -- Regards, Dmitry A. Kazakov http://www.dmitry-kazakov.de |