The first extension (in fact it is a longstanding feature
of Haskell) is to allow programmers to specify hierarchies
of classes. The class declaration shown here introduces
(a simplified version of) Haskell's Ord class, all of whose
instances can be compared using the (<=) operator. The
first line of the declaration specifies that every instance
of Ord must also be an instance of Eq. Think of the => symbol
as superset rather than implication in this particular case,
so it tells us that Eq is a superset of Ord. We say that Ord
is a subclass of Eq, or that Eq is a superclass of Ord.
A constraint like this makes sense because, if you can
compare two values using a (<=) operator, then there is
obviously a way of testing them for equality (i.e., by
taking (x==y) = (x<=y && y<=x)). In practice,
it means that we can often simplify deferred constraints
in a type, for example, by removing any constraint of the
form Eq a if we also have another constraint of the form Ord a.
It is the task of Haskell compilers/interpreters to ensure
that the instances declarations in a program are consistent
with the hierarchy of classes. For example, a program
containing an instance Ord T, for some type T, without also
providing an instance Eq T, would be rejected.