Package Semantics

Translation schemes One particularly systematic approach to development encodes a selection of patterns as translation schemes. To move from requirements spec to detailed code, you choose a succession of translations. Template packages can be used to represent translation schemes. Another variety of translation scheme expresses the semantics of a language, such as UML, by defining how it translates to more basic terms. This technique is especially useful for stereotypes, the variable part of the UML notation. 

To close full circle, as perhaps you may have guessed by now, dialects and semantic rules themselves are defined within packages, although, as we ve said, consider them virtual until further notice. But here is a short example to show the idea. 

Semantic rules are expressed as templates a dialect contains nested packages for its semantic rules. Each rule translates a slightly higher-level notation into its equivalent lower-level one. Here, any line between two type boxes that contains an explicit stereotype means the same as inverse attributes (see Figure 9.41). So what should an association line mean if it has no stereotype tag To define a default, you identify the untagged feature with the appropriate tag9 (see Figure 9.42). 

And, of course, most of these packages—especially the complex semantics ones—are virtual at present. But some research projects have indeed built up the packages of primitives. (The example given here comes from [Muial91].)... 

Join the action specifications and invariants from different views, as decribed in Chapter 9, Model Frameworks and Template Packages (p.371). The covariant join semantics permits different views to impose their own restrictions on the whole. 

OCL provides a fixed set of basic types and operations on numbers, sets, sequences, and so on. No such set will suffice for real applications unless they can be extended many common constraints would be awkward if limited to the OCL set. The Catalysis semantics of packages lets us extend these basic types and operations without requiring any change to OCL. Thus, Catalysis sequences could support something like this ... 

Unlike many designer-oriented specifications, data flow graphs are semantically clean and simple, thus forming an unambiguous behavioral definition suitable to interface to or exchange between synthesis packages, as well as to formal verification. 

---