Section 2 also stresses the notion that proof expressions are sensible for classical logic (see [30]). This separates an important technique from its origins in constructive logic. So the account of Typed Logic covers both constructive as well as classical logic. It also lays the ground work for tactics.
Section 3 features a very simple fragment of type theory which illustrates the major design issues. The fragment has a simple (predicative) inductive semantics. We keep extending it until the semantics requires the insights from [2].
The treatment of recursive types in Section is quite simple because it deals only with partial types. It suggests that just as the notion of program ``correctness'' can be nicely factored into partial and total correctness, the rules for data types can also be factored this way.
Table 1 shows how similar concepts change their appearance as they are situated in the three different contexts.