Open and efficient type switch for C++

Abstract

Selecting operations based on the run-time type of an object is key to many object-oriented and functional programming techniques. We present a technique for implementing open and efficient type switching on hierarchical extensible data types. The technique is general and copes well with C++ multiple inheritance.To simplify experimentation and gain realistic performance using production-quality compilers and tool chains, we implement a type switch construct as an ISO C++11 library, called Mach71 . This library-only implementation provides concise notation and outperforms the visitor design pattern, commonly used for case analysis on types in objectoriented programming. For closed sets of types, its performance roughly equals equivalent code in functional languages, such as OCaml and Haskell. The type-switching code is easier to use and is more expressive than hand-coded visitors are. The library is non-intrusive and circumvents most of the extensibility restrictions typical of the visitor design pattern. It was motivated by applications involving large, typed, abstract syntax trees.

Open pattern matching for C++

“…Our Match-statement extends the efficient type switch for C++ [3]. The core of this extension amounts to handling of multiple subjects (both polymorphic and non-polymorphic) as well as the ability to accept patterns in case clauses.…”

Open pattern matching for C++

“…Mach7 employs an efficient type switch [41] under the hood to convert subject type to accepted type. Guards, n+k patterns, the equivalence combinator, and potentially some new user-defined patterns depend on capturing the structure (term) of lazily-evaluated expressions.…”

“…In a library setting, we are able to provide a more efficient implementation if we keep the arguments separated. This is why our Match statement extends the efficient type switch for C++ [41] to handle multiple subjects (both polymorphic and non-polymorphic) ( §3.5.1) and to accept patterns in case clauses ( §3.5.2).…”

“…The core of our efficient type switch [41] is based on the fact that virtual table pointers (vtbl-pointers) uniquely identify subobjects in the object and are perfect for hashing. Open type switch maps these vtbl-pointers to jump targets and necessary this-pointer offsets and provides an amortized constant-time dispatch to the appropriate case clause.…”

“…the number of arguments N , the subject types subject typei (defined as decltype(ei) modulo type qualifiers), and the number of polymorphic arguments M . When M > 0 it also introduces the necessary data structures to implement efficient type switching [41]. Only the M arguments whose subject typei are polymorphic will be used for fast type switching.…”