--------------------------------------------------------------------------
Adaptive Object-Oriented Software Development	Fall 1997
COM 3360/NTU SE737				Prof. Karl Lieberherr
---------------------------------------------------------------------------
Project : Technology Transfer                   Submitted by: Karen Yahle

==================================================================

My original project was to be Object Equivalence of Class Graphs.
I had problems with that project and decided afer a few weeks
I needed to modify my selection of topic if I was to complete this
project.  

This technology transfer project was less complicated and I had
already solved it first in "C" and then again using C++.  The
growth plan for this project details the steps I used to build 
the functionality slowly.  

One of the reasons I chose this project was that in the original,
I had written a number of parsing functions to properly 
identify and handle the different types of input that needed
to be processed.  Demeter/Java's built-in parsing greatly
simplfied the processing of the input and the code that handled
it.  

This project simulated an insurance database with a list
of agents, a list of customers and their policies and
a list of transactions to be executed on the database.  

There are 6 types of transactions possible:

    1) this transaction will print all insurees in the database.
       input format: transaction1 1
       output  format: lastname, firstname.

    2) this transaction will find all insurance agents that have
       sold at least one policy of the same type to more than one
       customer.
       input format: transaction2 2
       output format: agent: LASTNAME, FIRSTNAME, AgentCode

    3) this transaction will find all agents that have sold the
       type of insurance specified in the query field.
       There are 3 types of policies possible: life, accident, home.
       input format: transaction3 3 accident
       output format: agent: LASTNAME, FIRSTNAME

    4) this transaction will find all agents who have sold a total
       greater than the amount in the query field.
       input format: transaction4 4 2200
       output  format: agent_lastname, total $$ premiums

    5) this transaction will find all insurees that have bought
       insurance from the agent specified in the query field.
       input format: transaction5 5 CLINTON, BILL
       output format: insuree: LAST, FIRST; agentCode: 123456

    6) this transaction will find all insurees who have bought more 
       than one type of insurance from the same agent.
       input format: transaction6 6
       output format: insuree: LASTNAME, FIRSTNAME.

   Example of inputs:
   "agent:" LASTNAME, FIRSTNAME, AgentCode

   "insuree:" LASTNAME, FIRSTNAME  "Policies:" "(" type, premium, agentcode ")"

   "transaction#" trans type [optional additional query data ]

Directory: files and subdirectories are attached to email message in zip format.  
           (see below)

Files for this project:  (contained in top directory)
    Class dictionary:  project.cd
    Behavior file: project.beh
    Input files: agent.input
                 insuree.input
		 trans.input

    Output file: sample output in project.output

    install.direct: this file has installation and execute directions
		    for this project (for alexey)
    README-proj: this file

    project.html: contains the tasks, growth plan, use cases,
		  results and summary for the project.

    /original: subdirectory containing the C and C++ programs used
	       to base this project on. 
   
    /vers2 thru /vers7:  these are subdirectories containing the 
	       different phases of development. (if you're interested)
   
Parts of your project which you would have developed further if
you had more time. Include a list of known bugs here.

  Functionality that I did not develop was the ability to print the
  output in ascending or descending order.  In the C/C++ programs this was
  accomplished with ordered doubly-linked lsts so that one could start
  at either end depending on which order was required.  Also, when an
  agent or insuree was repeated in the output more than once, it should
  not be printed again.  This feature was not implemented. 

  My use of Demeter/Java could very well have been more elegant and the
  code size smaller. I could possibly have reused traversals and visitors 
  using  alternation classes (ie. in the transaction classes).  I think
  I might have used in-lined visitors and traversals more often rather
  than defining them as I did.
  I also tried writing some "dig_out" methods so I did not have to reference
  subparts directly, but had some problems in getting to the correct
  data, so backed them out.  

Please answer the following questions: 

Did you change the generated Java code? NO

----------------------------------------