on December 29, 2002
This text is perhaps the most accessible yet thorough introduction to type systems I've encountered.
On the one hand, it offers excellent grounding: practical motivation is provided, numerous examples illustrate the concepts, and implementations are provided which can be used to typecheck and evaluate these examples. At various points, extended demonstrations of the type systems under consideration are given (e.g. showing how objects may be encoded). The exercises are well constructed and in many cases, accompanied with answers and detailed explanations in the appendix.
On the other hand, it offers an excellent exposition of the material: Pierce provides a lucid account of the static and dynamic semantics (primarily small-step operational) for various lambda calculi. He proceeds in a stepwise fashion via the gradual accretion of features: from first order (simply typed) systems to higher order systems incorporating bounded subtyping and recursion. He also gives attention to the metatheory of these systems (focusing on proofs of progress and preservation, and for systems with subtyping, of decideability). Internally, the text is well organized, with clear dependencies among the chapters, and the bibliography is extensive.
It should be noted that, while reasonably comprehensive, the text is necessarily limited in scope. For example, aside from the discussion on Featherweight Java, systems other than typed lambda calculus variants are not considered. In my opinion, the focus on these (in some sense "low-level") calculi makes foundational issues more apparent, and the linear progression from simple to complex variants lends a pleasant cohesiveness that would have been lost in a more general survey. However, as object/class encodings were discussed at various points, it would have been nice to see a more integrated presentation, in the spirit of the paper Comparing Object Encodings [BCP97].