MacCormackKemererCusumanoCrandall03

• Alan MacCormack & Chris F Kemerer & Michael Cusumano & Bill Crandall
• Trade-offs between Productivity and Quality in Selecting Software Development Practices
• IEEE Software Magazine V20n5(Sep/Oct 2003)pp78-85
• =POLL HP PROJECTS 2000-2001 TECHNIQUES QUALITIES STATISTICS ONE-SIZE
• Defect rate:= customer reported defects per LoC per month of use.
• Productivity:= LoC per month of development
• Significant Correlations:
  Net
  1. Systems projects had a larger defect rate.
  2. Larger systems had a lower defect rate and higher productivity.
  3. A complete functional specification increases productivity.
  4. Customers using a more incomplete prototype decreased the defect rate and productivity.
  5. Design reviews reduce defect rate.
  
  (End of Net)
  
• Best_fit models:
• defect_rate= 16+15*system+0.5*prototype_completeness-12*regression testing-20*use_reviews.
• productivity= 35-0..4*prototype_completeness+17*daily_builds.
• Conclusion: one size does not fit all. Choose coherent set of practices to meet desired qualities.

Seidewitz03

• Ed Seidewitz
• What Models Mean
• IEEE Software Magazine V20n5(Sep/Oct 2003)pp26-32
• =ESSAY MODEL SEMANTICS INTERPRETATION THEORY METAMODEL UML
• SUS:="System under study".
• Amodel is a set of statements about a SUS. It iscorrect if all the statements are true of the SUS.
• If a model acts as a specification of a SUS then the SUS is valid relative to the specification if all the statements are true.
• Aninterpretation is a mapping from model elements to the SUS. It come from a UML profile and/or local standards.
• If an interpretation is invertible the model is a representation of the SUS.
• A theory is a way to derive new true statements from a correct model.
• A model conforms to a theory if the theory only adds statements that are consistent with previous statements.
• If a model conforms to a theory and is interpreted by a SUS then the theory can derive new facts about the SUS.
• A modeling language expresses models for a class of SUS.
• A metamodel is a model that specifies a family of models for a given class of SUS.
• A reflexive metamodel describes a modelling language, using that modelling language.
• So meaning can mean interpretation or theory.

KulkarniReddy03

• Vinay Kulkarni & Sreedhar Reddy
• Separation of Concerns in Model-Driven Development
• IEEE Software Magazine V20n5(Sep/Oct 2003)pp64-69
• =DEMO METAMODEL TEMPLATE UML OCL
• Layered architecture implies many similar transformations.
• The Template meta-model can be instantiate to fit them all, including code generation of GUIs, functions, and databases.
• MetaModel

FranceGhoshSongKim03

• Robert France & Sudipto Ghosh & Eunjee Song & Dae-Kyoo Kim
• A Meta-modeling approach to pattern-based model refactoring
• IEEE Software Magazine V20n5(Sep/Oct 2003)pp52-58
• =DEMO PATTERNS MODEL METAMODEL REFACTOR
• examples: bridge, observer, abstract factory, GoF Maze Game
• Metamodel defines a set of source models and target models and the transformation between them.
• Tool with 2 user interfaces: pattern engineer(metamodel) and application engineer(model)

Bock03

• Conrad Bock
• UML without Pictures
• IEEE Software Magazine V20n5(Sep/Oct 2003)pp33-
• =DEMO MODEL REPOSITORY UML2.0 SEMANTICS COMPILATION ACTIVITY
• Shows how a C++ function is represented by an activity diagram and by collection of objects (mini data base)that model the diagram -- a repository

Selic03

• Bran Selic
• The Pragmatics of Model-Driven development
• IEEE Software Magazine V20n5(Sep/Oct 2003)pp19-25
• =ARTICLE MDD MODEL
• MDD needs: Automatic translation between models, and into/out of code. Standards.
• Models need abstraction, understandability, accuracy, predictiveness, and inexpensive.
• One must be able to execute models and relate code execution to models.
• Generated code must be efficient.
• MDD targeted at large-scale industrial applications.
• Need a fit with legacy systems and environments.