250 likes | 336 Views
Developing Engineered Product Support Applications. H. James Hoover, Tony Olekshy, Garry Froehlich, Paul Sorenson. Software Engineering Research Laboratory, Computing Science, University of Alberta www.cs.ualberta.ca/~serl. Avra Software Lab Inc., www.avrasoft.com.
E N D
Developing Engineered Product Support Applications H. James Hoover, Tony Olekshy, Garry Froehlich, Paul Sorenson Software Engineering Research Laboratory, Computing Science, University of Alberta www.cs.ualberta.ca/~serl Avra Software Lab Inc., www.avrasoft.com Work supported by Natural Sciences and Engineering Research Council of Canada and National Research Council of Canada V1.3 SPLC1 Talk
Outline • Engineered Product Domain • Common Application Product Line Requirements • Frameworks • Killer Best Practices • Conclusion SPLC1 Talk
Requirements Tools Product Our Context Software Builder Engineered Product Manufacturer Engineered Product Purchaser SPLC1 Talk
Application Domain Context Domain Variability Engineered Product Sizing and Selecting Engineer’s Requirements Ordering of Engineered Product Engineering Standards Tool Support Workflow Variability Platform Variability Product Catalog SPLC1 Talk
Business Process Context SPLC1 Talk
Application Requirements • Product Specific • Engineering Process • Generic Support • Uncertainty Everywhere SPLC1 Talk
Engineering Tool Manifesto Thou shalt be: Consistent Observable Verifiable Auditable Extensible SPLC1 Talk
Engineering Requirements • All tools have consistent behaviour and feel. How are: • values and associated units displayed and entered? • base units maintained within calculations? • changes brought to the attention of the user? Consistent Observable Verifiable Auditable Extensible SPLC1 Talk
Engineering Requirements • The tool should ensure that the user is aware of what: • has been completed • remains to be done • are effects of current action Consistent Observable Verifiable Auditable Extensible SPLC1 Talk
Engineering Requirements Consistent Observable Verifiable Auditable Extensible • Preserve internal & displayed values. • Trace calculation internally. • Use external program to verify. SPLC1 Talk
Engineering Requirements • Tools must be able to: • check their inputs for tampering or corruption, • produce outputs with the same properties. • Want equivalence to a stamped and signed drawing. Consistent Observable Verifiable Auditable Extensible SPLC1 Talk
Engineering Requirements Consistent Observable Verifiable Auditable Extensible • Make it possible for the engineer user to extend the tool. • But preserve all the preceding properties! SPLC1 Talk
Product Line Architecture • Our PLA is a set of domain specific application frameworks. • Each sub-framework provides a small set of services. • An application is a collection of services, with various degrees of coupling. SPLC1 Talk
User Agents Forms/Wizards Business Objects Domain Specific Services UI Manager Persistent Object Manager Foundation Services DB-Centric PLA SPLC1 Talk
Framework Design Goals • Make it easy to evolve the UI and workflow. • Preserve engineering integrity of the application under evolution. • Support deployment-related activities. • Anticipate refactoring of services between sub-frameworks. SPLC1 Talk
Killer Features • Engineering Features • Core Features • Persistence Features SPLC1 Talk
Engineering Features • Worksheet navigation • Worksheet language • Basis and Display Units • Special Input Widgets • Check Worksheets • Worksheet version migration • External Testing Hooks • Database Access Keys SPLC1 Talk
EAF Calculation Block A B Externals Inputs Outputs X, Y Z locals T = AX+BY Z = T + 2 T SPLC1 Talk
EAF Worksheet C: X X C.A C.B Y C.T D: D.A D.B Z U D.T SPLC1 Talk
Workflow SPLC1 Talk
Core Features • Trouble Stack • Data Signatures • Configuration Report • HTML Reports and Displays • Apalon markup language • Handbook writer’s assistance SPLC1 Talk
Persistence Features • Object Model, ID, Foreign Keys • Concurrent Transaction Consistency • Referential Integrety • Journalling, Roll-forward, Revision Control • Replication • Schema Conversion SPLC1 Talk
Conclusion • Didn’t develop all at once, important to support evolution. • Architecture is more important than implementation (thick => thin). • Framework embodies process and practices of domain and developer. SPLC1 Talk
Conjecture/Intuition • Architectures that survive variability over time survive variability over space. • We should be building a best practices catalog of reference architectures. • There may be no quantitative theory of architecture. SPLC1 Talk