Expression Compatibility Problem

Abstract

We solve the Expression Compatibility Problem (ECP) -a variation of the famous Expression Problem (EP) which, in addition to the classical EP concerns, takes into consideration the replacement, refinement, and borrowing of algebraic datatype (ADT) cases. ECP describes ADT cases as components and promotes ideas from Lightweight Family Polymorphism, Class Sharing, and Expression Families Problem (EFP). Our solution is based on a formal model for Component-Based Software Engineering that pertains to EP. We provide the syntax, static semantics, and dynamic semantics of our model. We also show that our model solves EFP as well. Moreover, we show how to embed the model in Scala.

Solving the Expression Problem in C++, á la LMS

“…In this paper, we use parts of the γΦC 0 calculus developed for solving the Expression Compatibility Problem [15]. γΦC 0 was developed after observing that sharing ADT cases amongst ADTs is not limited to ADTs only extending one another.…”

Instructed late binding

“…Using γ ΦC 0 [12] as the formalism, one denotes those as: NA = N um ⊕ Add and NAM = N um ⊕ Add ⊕ Mul. Notice how the latter formalism dismisses the subscripts for Num, Add, and Mul.…”

“…But, IDPaM is also inspired by Component-Based Software Engineering (CBSE) [23, §17], [19, §10]. Like previous EP solutions of Haeri and Schupp [11,8,12] to_string assembles building blocks provided by the programmer. (Cf.…”

Integration of a Decentralised Pattern Matching: Venue for a New Paradigm Intermarriage

“…Our technique demands coding discipline. Yet, whilst the weight of that is comparable for the component vendor to our former solution given in [15], the component user's job is now virtually automatic. Our technique works by integration of a decentralised pattern matching: Instead of centralising the pattern matching in a single place, we distribute it amongst components that correspond to the ADT cases.…”

“…It also makes ADT cases stand-alone entities -forming a fully fledged approach, manifested in [13,12]. In the accompanying formalism [15]: (i) an ADT case is represented using a component and is ranged over by γs; (ii) the ADT representation forms a family that simply combines components. Families are ranged over by Φs:…”