元建模级别
摘自[Tony Clark, Andy Evans, et al., Applied Metamodelling],<<>>内容为注释。
We are often asked by clients how they can assess the quality of a metamodel. To help them, we have found the following five levels to be useful:
Level 1 This is the lowest level.
A simple abstract syntax model must be defined, which has not been checked in a tool.
The semantics of the language it defines will be informal and incomplete and
there will be few, if any, well-formed rules.
the abstract syntax model will be relatively complete.
A significant number of well-formedness rules will have been defined, and
some or all of the model will have been checked in a tool.
Snapshots of the abstract syntax model will have been constructed and used to validate its correctness.
Some consideration will have been given to the extensibility of the language architecture, but it will not be formalised or tested.
the concrete syntax of the language will have been formalised and tested.
Users will be able to create models either visually and textually and check that they result in a valid instance of the abstract syntax model.
Models of semantics will have begun to appear.
All aspects of the language will have been modelled, including its semantics.
The semantic model will be executable, enabling users of the language to perform semantically rich operations on models written in the language, such as simulation, evaluation and execution.
self contained definition of the language that can be used ‘as is’ to generate or instantiate tools.
Most of the metamodels we have seen do not achieve a level greater than 2.
Even international standards such as UML do not exceed level 3.
Yet, reaching level 5 must be an aspiration for all language developers.
浙公网安备 33010602011771号