Linear typing schemes guarantee single-threadedness and so the soundness of in-place update with respect to a functional semantics. But linear

Syntactic Control of Interference (SCI) [Rey78] has long been studied as a basis for interferencefree programming, with cleaner reasoning properties and semantics than traditional imperative languages. This paper makes SCI-based languages more practical by introducing a revised version of Huang and Reddy's type reconstruction system [HR96] with two significant improvements. First, we eliminate the need for explicit coercion operators in terms. Although this can lead to more complex principal typings, we introduce some minor restrictions on the inference system to keep the typings manageable. Second, we consider adding let-bound polymorphism. Such a modification appears to be nontrivial. We propose a new approach which addresses issues of both polymorphism and interference control. Some examples show the utility of our system. SCI can be adapted to a wide variety of languages, and our techniques should be applicable to any such language with SCI-based interference control. 1 Introducti...