Classes as Invariants
intermediate20 min readLesson 73 of 204
Design order: invariant → members → operations; struct vs class without folklore.
A class bundles state and the operations that keep that state valid.
Beginner showed you class for encapsulation; now we use it as a design tool:
decide the invariant first, then expose only operations that preserve it.
#include <string>
class Timer {
public:
void start() { running_ = true; ++starts_; }
void stop() { running_ = false; }
bool running() const { return running_; }
int starts() const { return starts_; } // invariant: counts every start
private:
bool running_{false};
int starts_{0};
};
The invariant here is trivial ("starts_ counts calls to start"), but the
shape is what matters: state is private, operations are the only door,
const marks the read-only half of the interface.
Design order that works
- Write down the invariant in one sentence.
- Choose members that make the invariant easy to keep.
- Expose operations; make anything the invariant needs
private. - Mark non-mutating operations
const.
Struct vs class, honestly
The only difference is the default access (struct public, class private).
Use struct for passive data aggregates, class when there is an invariant
to defend. Both are "classes" in C++'s eyes.