• All ConstraintCollectors methods should be javadocced.
• Most can use our standard "As definied by ..." trick, toward the base one which is fully javadocced.
  • for countDistinct(), I think the base should be countDistinct(Function), not countDistinct()
• The base methods should use examples to avoid confusion. For example:
  • For example javadoc of count:

    from([Person("Ann", "Cole", 20), Person("Beth", "Cole", 20), Person("Carl", "Cole", 22), Person("Dan", "Jones", 20)]
    .groupBy(Person::getLastName, count()) returns [{"Cole", 3}, {"Jones", 1}]

  • For example javadoc of countDistinct:

    from([Person("Ann", "Cole", 20), Person("Beth", "Cole", 20), Person("Carl", "Cole", 30), Person("Dan", "Jones", 20)]
    .groupBy(Person::getLastName, countDistinct(Person::getAge)) returns [{"Cole", 2}, {"Jones", 1}]

• All examples should use our CS streams example conventions (the ones above might not, but I think they do). IIRC, that means similar to JSON: [] for collections, {} for objects, () for method calls.

See discussion: https://stackoverflow.com/questions/64695627/count-vs-countdistinct-in-optaplanner