Due:

Monday, February 24, 2020 by 5:00 pm.  For delivery see the Deliverables section below.

Purpose:

The purpose of this laboratory is to become comfortable with the continued use of our requirements and classification techniques as we turn our focus to writing code.  Building on Laboratory I, the purpose of this practicum is to become comfortable the exchange that takes place between coding and design.  The design revolves around the requirements for a university course registration system.  This type of system is highly instructive from a structural modeling standpoint, as there are several decisions that need to be made as regards inheritance, aggregation, and various associations between classes.  Make sure you understand the project requirements.

At this point, you should be thinking through in depth the requirement for Deliverable Six of Practicum I.  Practicum II and Practicum III all depend on the focused subset of overall requirements you will have chosen to implement.  Remember, you are not required to deliver the code for the entire system as specified in the General Description of the Problem Domain here.  Rather, you are to produce the code for a prototype that represents a solid subset of those requirements, and deliver code and tests for that prototype as your final project deliverable.  This practicum and the third practicum are all focused on that solid subset you have chosen.

Strategy:

1. Create a subdirectory called "Practicum2".

2. Create further subdirectories for each of the Deliverables below.  Note you may need to copy over files (say from your Visual Paradigm directory, or from your Eclipse directory, etc.) into these subdirectories.

Deliverable One:  Second iteration of your model, tests, and code (MTC) of your system.

As we have said, developing a solution in code for the entire system as specified in the requirements is not possible.  Update your Context Map (not your Summary Survey/Map) with any new Bounded Contexts have you have identified, if any.

Focus on a few of these bounded contexts you have identified in your High Level Architecture, among the bounded contexts you plan to delivery in your prototype.  Your decision to focus on one, two, or three or even more bounded contexts will depend on the contexts themselves and how much "work" you think will be involved in delivering the implementation for them.  Using tools from your modeling toolbox (Analysis Models, Context Map, High Level Architecture, CRC Cards, UML Class diagrams, etc.), clarify your work as you begin to actually write the code for your classes and the tests to ensure that your code is doing what you intend.  Experiment further trying to write tests prior to actually coding your classes.  Do not feel, however, as if your main deliverables here are models.  Your main deliverables here are code and tests.  However, when you get stuck in coding your classes, do leverage analysis methods and class modeling techniques to help you clarify your thinking.

As you begin your coding, you might want to start coding the classes for the tests your wrote in Practicum I. 

As you begin to write more code, try to do it strategically.  Is there a particular part of the model (a particular bounded context, for instance) that you think would be "low hanging fruit"...that is, easy to pull off quickly and easily?  Or you might want to focus on a particular class hierarchy or aggregate that you think you understand well (Instructor/TA/Student, e.g.).  You might want to start there.  Another approach might be to think of the part of the model that would yield the most "bang for the buck," that is, that would move you furthest along in the delivery of your system.  Only you can make these decisions as to where to start and what to continue.

Remember, your intent (evidenced in your deliverables) should be to demonstrate that you have THOUGHT DEEPLY about the problem and your software solution should be intellectually rich.  Your solution should demonstrate sound object-oriented design principles and patterns.  Your code and design should be S.O.L.I.D.  You will be graded on how well you COMMUNICATE that deep thinking through your models, tests and code (MTC).

As for details in your models, I do NOT expect a thoroughly detailed class diagrams for every class you plan to implement in your prototype.  I certainly don't expect ALL constructors, accessors, modifiers, and arguments for EVERY method for EVERY class.

Having said that, you should make sure you have thought through the problem well enough so that you have all of the salient classes you will need and their supporting classes, and have defined the interface on those classes sufficiently well for a reader of your model to determine from the class diagram what responsibilities are being handled by the class.  You should comment your code so that we understand the responsibilities of the class.  Model adornments such as stereotypes, etc., should be used when necessary to clarify your intent.  When it comes to coding, stubs and mocks are your friends.

Someone will ask the question:  "How many classes do we have to code for this deliverable?"  Our answer is that the final project is due in 4-5 weeks, so you should expect to deliver about 40-50% of the classes you anticipate needing for the final project in this deliverable.  It's a fair amount of work.  That said, we are certainly not going to count how many classes you have delivered.

You may find the following online resources helpful as you work together to produce these deliverables:

What is an Essential Use Case?

Domain-Driven Design Reference

Domain-Driven Design Quickly

GitHub References Points on Domain-Driven Design

http://guide.agilealliance.org/guide/crc.html

http://agilemodeling.com/artifacts/crcModel.htm

http://css.dzone.com/articles/crc-cards-primer


Deliverable Two:  Become comfortable with dynamic modeling

You are to look at the classes you have defined in your class model and code, as well as their relationships, and now turn your focus to fleshing out the operations required on those classes, by concentrating on the messages that will be passed among classes in their collaborations in runtime scenarios.  Define clearly the responsibilities of each class that you have, and decide if further abstraction (i.e., new classes) is necessary (it almost certainly will be). Think about single responsibilities.

To guide you in this endeavor, think through the various collaborations that objects will engage in during run-time, especially the sequence of these communications.  Use UML sequence and (optionally) collaboration diagrams to help your thought process on this.  The activity will be a recursive effort at identifying sequences of interactions and at the same time identifying new operations that will be needed to support the various collaborations.

Think about public versus private interfaces.  Public methods should only appear on your class definitions when you intend those methods to be part of the public interface of your class, that is, part of the interface your client objects will be dependent upon.  Everything else should be non-public.

You will undoubtedly discover that you need new classes (with smaller scope) to facilitate these object-object interactions.  You should begin to think of introducing Interfaces and/or Abstract classes strategically (cf. interface segregation) if you find parts of your code are too dependent on concrete classes (don't go crazy here, but hopefully you will discover some opportunities to make your life easier). You may find it useful to develop helper classes to control the various system activities and object interactions. Of particular use here would be Abstract Factories for object creation.

As you revisit your models and code, think of individual patterns that you have learned that might be applicable in a given context, with a view to simplifying either the static class structure or dynamic interactions among objects.  The design patterns you should have in your arsenal by the delivery date include Singleton, Class and Object Adapters, Template Method, Mediator, Observer, Abstract Factory, Composite, Iterator and State, Proxy, and Visitor.  Not a small arsenal.

You should create several UML sequence, collaboration, state, and activity diagrams which detail your understanding of how the objects based on your current model will interact and collaborate at run-time.  These diagrams should show your interpretation of the dynamic behavior of your model at run-time.

Your grade for this deliverable will be determined on the depth of insight you have brought to the creation of a dynamic representation of your static model.

You may find the following online resources helpful as you work together to produce these deliverables:

Sparx Systems help on Dynamic Modeling
TutorialsPoint on Dynamic Modeling
Excellent reminder of core concepts in Dynamic Modeling (Youtube)
Good overview on UML diagrams

Deliverable Three:  Begin to integrate MySQL into your solution.

Deliverables

The deliverables for each iteration should be placed in the appropriate sub-directory of the pre-existing "practicum2" directory inside your repo. Submit everything using Subversion before the date and time due.