Let's take the example of a dog veering toward shore or to a point on the way to a blind. Another example might be running around a pond. One method is to allow them to get right up on land (or around the pond) and then yell "no" and correct them.. This would be trying to stop the undesireable behaviour and adding an aversive-thus Positive punishment. It can act to create a "hot-spot" and in the future dogs would not go there. This used to be a common technique. Two problems: Yelling NO does not tell the do what to do. Secondly, the no nick can create an unbalanced dog. Today's dog has to be as comfortable on land as the water. They cannot be afraid to land. What they have to learn is how to fight factors that prevent them from going straight and losing their destination. They have to learn how to make good decisions.
The way I train is to watch the dog and determine the instant he makes a decision to deviate. That is the best time to intervene in his mind and thinking. I don't want to condone a bad decision and then Bam he gets it. Instead, I will blow a sit whistle and literal cast to the bird. Now if he deviates I have told him what to do and he has not made an effort to change. Now I can either repeat that sequence(attrition) or I can blow the sit whistle and give a nick. The nick is on something he knows(sit) and it is negative reinforcement of the sit. This aversive reinforcement acts indirectly to promote compliance on the cast. We call that Indirect pressure.
If you balance your training and make your interventions clear you lessen having dogs afarid to get on points. My dog's default is often to swim by the point given a choice but they should get on the point if I ask. Of course, some days they seek the point and sometimes they refuse my asking. I work on this forever and design 3-peats and tune-ups and cheating marks to keep these skills as sharp as possible.
The theory guides the procedure. Consistent application and persistent effort guide the practice.