Type declaration

With reference to the algebraic specification of data types, the specification of a type T, defines a set (usually infinite) of constants (T-constants). Besides, certain T-specific properties (T-axioms) are valid for these constants (semantic of T). For example, the type NATURAL defines the set N=(0, 1, 2, 3, 4, 5. ..) of NATURAL constants, where the relation 2=SUC(SUC(0)) is one of the T-axioms (SUC denotes the successor function / SUC(X) =X+1). 

The HDSL knows three different type classes 

Syntax type declaration = ENTITY type name = type  

STRUCTURE attr1 REAL attr2 LOGICAL attr3 NAME END  

---

Type declaration—Specifies data type to be represented by a variable name (overrides defaults), e.g.,... 

Data Types (Data structures of ail types, predefined and user-defined, must be declared see Statements for type declaration form.)... 

Type declaration—Must be placed after CONST sequence and before variable declarations used to define nev (not predefined) data types either as a range of values or as specific values within parentheses and to define arrays, records, and files. User-defined scalar types are ordered in increasing lefl to right sequence. For example. 

Therefore, define a type (that is, an interface or pure abstract class plus a specification) that characterizes only the features that you will use. Declare the variable to be of this type. Declare the object s class to be an implementor of this type. This approach will minimize the dependency of your component on the other object. Related patterns are adapter and bridge [Gamma94],... 

Variables (reg and integer types) declared locally within an always statement do not infer flip-flops. This may potentially lead to a functional mismatch between the Verilog HDL model and the synthesized netlist. Here is an example of a locally declared variable Temp that does not get inferred as a flip-flop. 

The type-declaration characters (%,, , , ) cannot be embedded in a name. 

You can also declare a variable s type by appending a type-declaration character to the variable name, a technique from older versions of BASIC. The type-declaration characters include % for integer variables and for string variables. 

Besides these, most DBMSs provide interfacing mechanisms with which developers can implement applications in a general-purpose language (referred to in this context as a host language) and use the latter to invoke whatever operations are supported over the stored data (because they are part of the behavior of either the data types made available by the data model or the application types declared in the application schema). 

Flows in AADL describe the different sequences of an information flow through a set of contributing components. The description of this flow is subsequently used in certain analyses such as a flow latency analysis. In AADL, flows are defined with a flow specification and a flow implementation. A flow specification represents the externally visible flow of information in a component it is specified within the component type declarations using flow sources, flow paths and flow sinks (Feiler Hansson,2007). A flow source represents the originator of the flow, the flow sink represents the end consumer of the flow information, and the flow p>ath embodies the link between incoming and outgoing ports involved in the flow. A flow implementation on the other hand represents the actual realization of a flow within a component it is specified within the component implementation declarations. 

Each model can have parameters, variables, equations, initial conditions, boundary conditions and submodels that can have submodels themselves. Models can be based in pre-existing ones, and extra-functionality (new parameters, variables, equations, etc.) can be added. So, composition hierarchical modelling) and inheritance are supported. Every parameter and variable in a model is based in a predefined type and have a set of properties like a line/description, lower and upper bounds, unit of measurement among others. As models, types can have subtypes and the object-oriented paradigm is implemented. Some examples of types declarations can be seen in Figure 2. 

Definition 4-1 A specification of a procedure for predicate r//i, denoted Spec r), is composed of a procedure declaration, a type declaration for each parameter, a set of restrictions on parameters, a set of directionalities, and a relation definition. 

Uses a version of Pascal, extended to support type declarations (including bit widths), register declarations, and port declarations. 

A constant is an object with a static value of a given type, declared... 

Access and file types are not supported in the s5mthesis environment and will not be discussed. AdditionaUy a new type must always be declared in full, immediately - incomplete type declarations are ignored by the synthesizer. 

The enumeration literal can be an identifier (e.g. RED, Blue, green) or a character literal (e.g. V, X, ). Eadi must be unique within the type declaration. Identifiers (simple names) are discussed in detail in Chapter 4. 

The same enumeration literal can appear in more than one type declaration. In such cases it may be necessary to clarify the e being used in a particular context. A qualified expression can be used for this purpose. 

An array can be declared to be either constrained or unconstrained. A constrained array type has its range indices specified when the type is declared. All objects of this type then declared will have the same length. An unconstrained array type does not have the range indices fixed by the type declaration stage. Instead, they are only specified when an object of that type is declared. This means ihat objects of an unconstrained array type may all be different lengths. A two-dimensional array may have one or both pairs of indices initially unconstrained. Range indices must be static. 

BIT VECTOR is an example of a one-dimensional, unconstrained array type. In this and all array type declarations, three parameters must be specified - the type name, the range indices and the type of the array elements. In an unconstrained type, the only part of the range indices that is specified is their type. For example, the indices supplied when defining an object of type BIT VECTOR must be integers, greater than or equal to 0 - the definition of the t) e NATURAL (section 3.3.3). The box symbol <> indicates that the range is presently unspecified. 

Common items that are found in packages include type and subtype declarations, functions and components. Types and subtypes that are implicitly accessible to any design are declared in the STANDARD package. This package cannot be modified by the user and is stored in die STD library along with another, TEXTIO. However, the type declarations and procedures that this contains cannot be used if a design is to be synthesized. 

This ardiitecture also illustrates the use of Selected names to access libraries and packages. The ARITH library contains the compiled package ARITH TYPES and to access the SIGNED type declaration the Library and Use clauses have been used in the normal way. To access the operator function a Selected name has been used. This directly selects the library, package and function without the need for a Use clause. Section 3.4.3 discussed accessing package items in this way. 

The third architecture RIPPLES takes this direct access a step further and dispenses with the Use clause altogether. Both the SIGNED type declaration and the RIPPLE function are accessed directly. This architecture accesses the second RIPPLE function, overloading the first by the number of parameters that are supplied. This is only one way of achieving function overloading Box 6.8 discusses these methods further. Another approach selects the appropriate function by the type of the parameters that are supplied. 

As the above example demonstrates, ambiguous expressions that need qualifying usually occur because the same enumeration literal(s) occurs in more than one type declaration. 

The difference between resolved and unresolved types is discussed in Box 7.7. Of the two types declared in the STD LOGIC package, stdjogic is the one from which a variety of subt)q es are declared and sets the main standard. As already mentioned, the XOlZ is a subtype of this type, but others such as UXOlZ also exist. These can exploit the wide range of functions available for the type and at the same time maintain their portability. 

The multivalued logic systems outlined above are subsets and supersets of other systems. However, as the discussion in Chapter 3 on types and subtypes explained, objects of different types caimotbe university interchanged or equated. For example, given the type declarations ... 

The elements of a record must be accessed using a selected name. A selected name references an element of a record by expanding the name. Consider the type declaration below. 

Syntax = ENTITY <type name> = ... 

---