SlideShare a Scribd company logo

Towards More Flexible Enterprise Information Systems

•
•
CONFENIS 2012
CONFENIS 2012

This document discusses an approach called EIS Patterns for developing enterprise information systems in a more flexible way. EIS Patterns uses Python and Django along with techniques like test-driven development, decorators, and natural language processing to implement business processes. Decorators allow behaviors to be mixed and matched at runtime. Contracts between nodes and decorators are defined using a domain-specific language called Should-dsl and tested automatically. The goal is to have a transparent and configurable workflow engine that can be used with existing platforms and databases.

1 of 11
Download to read offline
Rogerio Atem de Carvalho, Instituto Federal Fluminense Bjorn Johansson, Lund University Towards More Flexible EIS
Introduction ● Flexibility: capacity of being adaptable in a cost and effort effective way ● Flexibility through relational databases based configuration is limited ● EIS Patterns: a testbed for testing new techniques for developing EIS ● Based on ideas brought from 10 years of participation in the open source ERP5 project ● Focused on the software development aspects
Environment ● Python, high level, dynamic programming language ● Automated testing: ● Test Driven Development (TDD) ● Behavior Driven Development (BDD) ● On the way: Business Language Driven Development (BLDD) ● Django Web development framework
Structure
Implementing Processes ● Business processes implementation: “a Process coordinates Nodes to perform Operations on Work Items” ● “a Movement encapsulates an Operation”: ● An operation represents a production task ● When trigged by an event in a Process, it defers the actual execution to a pre- configured node's method ● This execution is logged by a movement object, which stores all parameters, date and time, and results of this execution
Implementing Processes ● Therefore, an operation is an abstract concept which can be configured to defer execution to different methods, from different objects, in accordance to the intents of a specific business process instance ● Keeps business processes abstraction, while customization can be obtained by configuration
Decorators ● Subclassing avoidance: ● Subclassing adds behavior to all instances of the original class, decoration can provide new behavior, at runtime, for individual objects: “pay-as-you-go” approach to adding responsibilities ● Decorators allows mix-and-matching of responsibilities ● Decorators facilitates system configuration ● Problem: typically, it is necessary to deal with lots of small objects
Decorators and Contracts ● In order to decorate a Node, it is necesssary to check if this node is compatible with the decorator ● Should-dsl, a language created for writing more readable tests is used to check contracts ● Contracts are tested automatically, and can also be queried ● An example rule for given someone manager responsibilities: candidate_node |should| be_decorated_by(EmployeeDecorator)
Natural Language Processing ● Used to check requirements written in text ● Significant terms are used to parse decorators' documentation ● Operations are special methods that can be cathegorized, for instance as, financial, logistics, manufacturing... ● After finding candidate decorators, rules written in should-dsl are ran to check the compatibility between them and candidate nodes
Inoculable Workflow Engine ● WfE usally are of two types: ● App development platforms: configurable, but you must use the platform ● Software libraries: you can use your platform, but you need to program and use the library's objects ● xFluidity: ● Inoculable and expellable: transparent ● Configurable – even at runtime
Conclusions ● EIS Patterns is a testbed for: ● Automated testing techniques ● Dynamic programming languages ● Use of Domain Specific Languages (DSL) for testing, designing, and checking contracts ● Use of decorators for extenting and customizing behavior ● Use of NLP for requirements analysis ● Use of a transparent workflow engine ● Works with relational databases

More Related Content

Towards More Flexible Enterprise Information Systems

  • 1. Rogerio Atem de Carvalho, Instituto Federal Fluminense Bjorn Johansson, Lund University Towards More Flexible EIS
  • 2. Introduction ● Flexibility: capacity of being adaptable in a cost and effort effective way ● Flexibility through relational databases based configuration is limited ● EIS Patterns: a testbed for testing new techniques for developing EIS ● Based on ideas brought from 10 years of participation in the open source ERP5 project ● Focused on the software development aspects
  • 3. Environment ● Python, high level, dynamic programming language ● Automated testing: ● Test Driven Development (TDD) ● Behavior Driven Development (BDD) ● On the way: Business Language Driven Development (BLDD) ● Django Web development framework
  • 5. Implementing Processes ● Business processes implementation: “a Process coordinates Nodes to perform Operations on Work Items” ● “a Movement encapsulates an Operation”: ● An operation represents a production task ● When trigged by an event in a Process, it defers the actual execution to a pre- configured node's method ● This execution is logged by a movement object, which stores all parameters, date and time, and results of this execution
  • 6. Implementing Processes ● Therefore, an operation is an abstract concept which can be configured to defer execution to different methods, from different objects, in accordance to the intents of a specific business process instance ● Keeps business processes abstraction, while customization can be obtained by configuration
  • 7. Decorators ● Subclassing avoidance: ● Subclassing adds behavior to all instances of the original class, decoration can provide new behavior, at runtime, for individual objects: “pay-as-you-go” approach to adding responsibilities ● Decorators allows mix-and-matching of responsibilities ● Decorators facilitates system configuration ● Problem: typically, it is necessary to deal with lots of small objects
  • 8. Decorators and Contracts ● In order to decorate a Node, it is necesssary to check if this node is compatible with the decorator ● Should-dsl, a language created for writing more readable tests is used to check contracts ● Contracts are tested automatically, and can also be queried ● An example rule for given someone manager responsibilities: candidate_node |should| be_decorated_by(EmployeeDecorator)
  • 9. Natural Language Processing ● Used to check requirements written in text ● Significant terms are used to parse decorators' documentation ● Operations are special methods that can be cathegorized, for instance as, financial, logistics, manufacturing... ● After finding candidate decorators, rules written in should-dsl are ran to check the compatibility between them and candidate nodes
  • 10. Inoculable Workflow Engine ● WfE usally are of two types: ● App development platforms: configurable, but you must use the platform ● Software libraries: you can use your platform, but you need to program and use the library's objects ● xFluidity: ● Inoculable and expellable: transparent ● Configurable – even at runtime
  • 11. Conclusions ● EIS Patterns is a testbed for: ● Automated testing techniques ● Dynamic programming languages ● Use of Domain Specific Languages (DSL) for testing, designing, and checking contracts ● Use of decorators for extenting and customizing behavior ● Use of NLP for requirements analysis ● Use of a transparent workflow engine ● Works with relational databases