blob: 226bac50336d84d56297ffa939ef251540205dd4 [file] [log] [blame]
h1(#PivotProgrammersGuide). Pivot Programmers Guide
The Pivot Programmers Guide describes the ways in which the Pivot binding Eclipse OCL can be used from Java programs.
The Pivot binding was first available as an examples quality prototype in 3.1.0 (Indigo). The Pivot binding
will become the preferred binding in 3.3.0 (Kepler). The older Ecore and UML bindings are described in a separate "Ecore/UML Programmers Guide":#ProgrammersGuide.
The OCL Parser/Interpreter provides an implementation of the "Object Constraint Language 2.3":http://www.omg.org/spec/OCL/2.3 specification for EMF-based metamodels and models. It offers OCL
constraint and query parsing and evaluation, model-based validation, and
provides an infrastructure for content assist in textual editors.
The following features are supported in the current version:
* Classifier invariant constraints
* Operation precondition and postcondition constraints and body conditions
* Property constraints (initial-value and derivation)
* Attribute and operation definitions (def: expressions)
* Package context declaration
* Basic values and types
* Collection types
* Navigation of attributes and association ends
* Operation invocation
* Iteration expressions (all standard iterators)
* Let expressions
* If expressions
* Tuples
* Message expressions, including unspecified values
* Operations predefined by OCL: allInstances(), oclIsKindOf(), oclIsTypeOf(), oclAsType(), oclIsNew()
* Escape syntax for illegal names: type, operation, attribute, etc. names that correspond to OCL reserved words can be escaped in the standard fashion using a leading underscore ('_'). In addition, names that contain spaces or tabs can be escaped by enclosing them in double-quotes ('"'; this is non-standard). e.g., @self.ownedRule->forAll(c : Constraint | c._context = self)@
The above constructs are supported by the parser for parsing and
for evaluation, with the exception of the oclIsNew() operation and
message expressions. All of the above are supported for both Ecore
and UML models. The following are supported by default for UML
(both in parsing and evaluation):
* Navigation of non-navigable association ends (including those that are owned by the association)
* Qualified association end navigation
* Navigation to association classes, including source qualifiers
* Operations predefined by OCL: oclIsInState()
The following features are provided in addition to the OCL specification:
* String case conversion operations: toUpper(), toLower()
* Support for comparison (<, <=, etc.) and sorting of any java @Comparable@ s of conformant types
* Transitive closure of associations: closure(expr : OCLExpression) iterator
* Navigation of "hidden" opposites of references specified in Ecore models using a @Property.oppositeRoleName@ annotation with source @http://schema.omg.org/spec/MOF/2.0/emof.xml@ on the forward reference, producing an @OppositePropertyCallExp@ expression
The OCL implementation is defined in plug-ins for convenient deployment in
Eclipse, but as is the case for EMF, it can also be used stand-alone. The
plug-ins are partitioned thus:
* @org.eclipse.ocl@: the core parsing, evaluation, and content assist services. Definition of the OCL Abstract Syntax Model and Environment API. These APIs are generic, independent of any particular metamodel (though using Ecore/EMF as the meta-meta-model).
* @org.eclipse.ocl.ecore@: implementation of the Ecore metamodel environment, binding the generic Environment and AST APIs to the Ecore language. Provides support for working with OCL constraints and queries targeting Ecore models.
* @org.eclipse.ocl.uml@: implementation of the UML metamodel environment, binding the generic Environment and AST APIs to the UML language. Provides support for working with OCL targeting UML models.
Please refer to the "OCL Interpreter Tutorial":#OCLInterpreterTutorial for review of the code samples.