70 likes | 205 Views
Survey of Common Types I. Records E.g., structs in C++ If elements are named, a record is projected into its fields (e.g., via struct member selection operator as in P.x in C++) Variant records (e.g., unions in C++) let different types overlap in memory Arrays
E N D
Survey of Common Types I • Records • E.g., structs in C++ • If elements are named, a record is projected into its fields (e.g., via struct member selection operator as in P.x in C++) • Variant records (e.g., unions in C++) let different types overlap in memory • Arrays • Usually contiguous in memory and of a homogeneous type • Layout and access patterns also may affect cache effects, etc. • Provide an indexing operator (e.g., []) to access elements • May be allocated on stack or heap (language specific) • May be compatible with other types (e.g., pointers in C++)
Survey of Common Types II • “Utility” types (e.g., strings, sets, lists, files) • Widely used but not necessarily inherent to a language itself • Sometimes built into the language (e.g., Pascal) • E.g., set operators (+, *, -) for union, intersection, and difference • Sometimes provided by a library (e.g., the C++ STL) • Strings • Often just an array of characters (e.g., C style char *) • Often an object with operators (e.g., C++ style string) • Sets (similarly, lists) • Contain unique values (or in the case of lists, ordered sequences) of a common type • Files • “Streams” of data coming from and/or going to mass storage
Survey of Common Types III • Recursive types • May contain (an alias for) another instance of the same type • E.g., a list node contains an alias for the next node in the list • May require a pointer or other similar abstraction • To halt the nesting (and/or construction) of recursive type instances • Pointers and references • Similar in some languages, different in others • E.g., pointer/references semantics in Java vs. C++ • Also may be similar to other types • E.g., (const) pointer and array semantics in C++ • E.g., address arithmetic vs. index arithmetic in C++
More about Type Constructors • Used to construct sets from other sets • E.g., as Cartesian products (tuples) • If elements are named, a record is projected into its fields (e.g., via struct member selection operator as in P.x in C++) • Specialized forms of type constructors are common • Unions often are at best unnecessary in object-oriented languages (or worse: in C++ certain examples can exhibit type problems similar to those seen with reinterpret casting) • Subset type constructors narrow a set (e.g., in Ada to construct a sub-range type from a range type) • Array constructors project an index type into another type • Pointer/reference constructors create aliases for other types (pointers have their own operators but references do not)
Type Checking with Polymorphism • Hindley-Milner type checking • Strongly typed functional languages (ML and Haskell) do this • Unambiguous (deductive) type inference, by propagating constraints and info up and down an abstract parse tree • Relies on specific forms of instantiation and unification • Different types of polymorphism • Parametric (interface) polymorphism • Explicit subtype (inheritance) polymorphism • Liskov substitution principle: whenever S is a subtype of T, wherever you need an instance of T you can use one of S • Occurs-check in unification • Makes sure a variable isn’t in an expression substituted for it • Difficult to do efficiently, only use if necessary (it is for H-M)
Explicit Polymorphism of Data • Relies on syntax to declare forms of polymorphism that cannot be handled implicitly as in previous cases • E.g., declaring polymorphic data structures • Essentially, they provide user-defined type constructors • E.g., in C/C++ function pointer and array typedefs • Explicit parametric polymorphism • Explicit declaration of type parameters (e.g., C++ templates) can be used to make data types generic over other types • In C++ calls to function templates are resolved implicitly but template objects require concrete type parameter declaration • Expressions within parameterized functions/methods impose constraints on the parameterized types (concept/models relation in generic programming, see Austern on C++ STL)
Today’s Studio Exercises • We’ll code up ideas from Scott Chapter 7.3-7.9, 7.11 • Continuing our exploration of data types from last time • Today’s exercises are again in C++ • C++ allows us to explore many type system issues • Please take advantage of the on-line tutorial and reference manual pages that are linked on the course web site • As always, please ask us for help as needed • When done, email your answers with “Data Types Studio II” in the subject line, to the course account • Send to cse425@seas.wustl.edu