Manifest domains:analysis and description

Dines Bjørner*

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

304 Downloads (Pure)

Abstract

We show that manifest domains, an understanding of which are a prerequisite for software requirements prescriptions, can be precisely described: narrated and formalised. We show that such manifest domains can be understood as a collection of endurant, that is, basically spatial entities: parts, components and materials, and perdurant, that is, basically temporal entities: actions, events and behaviours. We show that parts can be modeled in terms of external qualities whether: atomic or composite parts, having internal qualities: unique identifications, mereologies, which model relations between parts, and attributes. We show that the manifest domain analysis endeavour can be supported by a calculus of manifest domain analysis prompts: is_entity, is_endurant, is_perdurant, is_part, is_component, is_material, is_atomic, is_composite, has_components, has_materials, has_concrete_type, attribute_names, is_stationary, etcetera; and show how the manifest domain description endeavour can be supported by a calculus of manifest domain description prompts: observe_part_sorts, observe_part_type, observe_components, observe_materials, observe_unique_identifier, observe_mereology, observe_attributes. We show how to model attributes, essentially following Michael Jackson (Software requirements & specifications: a lexicon of practice, principles and prejudices. ACM Press, Addison-Wesley, Reading, 1995), but with a twist: The attribute model introduces the attribute analysis prompts is_static_attribute, is_dynamic_attribute, is_inert_attribute, is_reactive_attribute, is_active_attribute, is_autonomous_attribute, is_biddable_attribute and is_programmable_attribute. The twist suggests ways of modeling “access” to the values of these kinds of attributes: the static attributes by simply “copying” them, once, the reactive and programmable attributes by “carrying” them as function parameters whose values are kept always updated, and the remaining, the external_attributes, by inquiring, when needed, as to their value, as if they were always offered on CSP-like channels (Hoare, Communicating sequential processes. C.A.R. Hoare series in computer science. Prentice-Hall International, London, 2004). We show how to model essential aspects of perdurants in terms of their signatures based on the concepts of endurants. And we show how one can “compile” descriptions of endurant parts into descriptions of perdurant behaviours. We do not show prompt calculi for perdurants. The above contributions express a method with principles, techniques and tools for constructing domain descriptions. It is important to realise that we do not wish to nor claim that the method can describe all that it is interesting to know about domains.

Original languageEnglish
JournalFormal Aspects of Computing
Volume29
Issue number2
Pages (from-to)175-225
Number of pages51
ISSN0934-5043
DOIs
Publication statusPublished - 1 Mar 2017

Keywords

  • Analysis & description
  • Domain engineering
  • Manifest domains
  • Prompt calculi

Fingerprint

Dive into the research topics of 'Manifest domains:analysis and description'. Together they form a unique fingerprint.

Cite this