Saturday, December 1, 2007

ProtoSOA

I'm thinking right now I wish I had more time to keep my blog updated!
At any rate, the current happenings on my plate are concerned with a new software product a colleague and I have been designing called "ProtoSOA". ComFrame Software is positioned to expand this design into a working product.

ProtoSOA is a Dynamic Enterprise Service Framework for Service Oriented Architecture (SOA) enablement. It is designed to allow business architects to define standard web services through metadata using administrative tools. The metadata not only provides the service descriptions, but also provides semantic mappings between canonical business terminology and “system terms” which are embedded in code. ProtoSOA not only exposes services that are fully compliant with current web service interoperability standards, it also seamlessly interacts with business process orchestration tools and provides the data transformation, transport, versioning, reliable messaging, security, auditing, and integration capabilities expected for an enterprise-class service bus.

If you would like to find out more about this product, please respond to this post.

That's all I have time for... later.

Wednesday, October 24, 2007

UML Training

Today I am preparing to teach a UML course at Emageon in Milwaukee, WI. I have presented this course several times for different clients in the past and have refined it each time. The central theme in my training is a "use case driven" process coupled with visual modeling. These two concepts are cornerstones to the USDP.

UML, therefore, is a visual expression and expansion of the use case model. The use case model is a reflection of the system requirements, and the UML model expands upon this model. It is beneficial to achieve both behavioral and structural views of the use case model using UML. Structural views manifest themselves as class diagrams, component diagrams, and collaboration diagrams, while behavioral views are manifested by activity diagrams, sequence diagrams, and state machines. Additionally, it is helpful to have a good mix of these diagram types -- each one is intended to address a specific viewpoint or level of detail.

Structural models address the "static" view of things. For example, class diagrams express the structure of classes, their associations with other classes, and their inheritance hierarchy. Behavioral models address the dynamic interaction of things in the system. For example, a sequence diagram emphasizes the messages between objects in a scenario, especially the sequence of those messages. It is very important to address both the structure and the behavior of objects in a system.

Another useful view is the State Machine, which is primarily used to express the complex behavior exhibited by a class, especially when the class exhibits different behavior depending on its state.

I have found that the most useful progression of UML modeling is as follows:
1) Begin with the use case model.
2) Expand the use case model using Activity diagrams.
3) Identify the "things" in the system. These are most evident by looking for nouns in the use cases and activity model. These are classes, modeled (of course) by a Class diagram.
4) Examine the simple behaviors of each class and how each class collaborates with other classes. Document these behaviors as operations on each class (in the Class diagram) and as collaborations (in the Collaboration diagram).
5) Document each scenario in which message sequence is critical by using a Sequence diagram.
6) In cases where objects exhibit complex, state-dependent behavior, use a State Machine.

For completed systems, use the Component and Deployment diagrams to depict the physical implementation and relationships among nodes.

Thursday, October 11, 2007

It occurs to me that some folks know about the Unified Software Development Process (USDP) and how it can ensure quality results. They may even act as process evangelists, helping IT shops implement process changes and USDP development techniques.

It also seems to me that some of them don't really believe in it. Maybe they understand the benefits from an academic perspective, but having never actually lived in a "process pure" environment, they see software development processes as intrusive and weighty.

The truth is, software development is chaotic by nature. The questions of "how do we begin", "how do we know when we are done", "what are our deliverables", and "what is quality" seem basic. However, it is surprising how often development projects are undertaken without understanding what these things mean. Even when process evangelists are involved from the beginning, when the team fails to follow the process systematically the project will begin to degrade toward chaos.

A good development process does not have to be heavy. The most important features (use case driven development, visual modeling, and iterative development) can be tailored to the project and the project team.

Following a USDP can lend order to the chaos. It provides focus to the development effort. Everyone on the team has a common mental picture of what must be done, how it should be done, what is being produced, how the components relate to each other, and how to know when they are done and done correctly. Even when things start to degrade, as they often do when schedules and budgets get tight, a good development process is self-healing. It provides the disciplines needed to regroup and re-focus the team, mitigate risks, resolve unforeseen issues, and establish corrective measures.

Actually putting the USDP into practice is the key to understanding its benefits. Once you have been on a development team whose focus is USDP, you will never want to do things the "old way" again.

Friday, September 28, 2007

My GNU Public Key

To encrypt data you send to me, or to verify docs digitally signed by me, please use my GNU Public Key:
-----BEGIN PGP PUBLIC KEY BLOCK-----
Version: GnuPG v1.2.2 (MingW32)
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=/RXh
-----END PGP PUBLIC KEY BLOCK-----

Wednesday, September 12, 2007

Let's begin...
I really want to share with you some thoughts about software development.
These are some "truths" that keep making their way into my experiences. Perhaps you have seen them too.
1) Keep it simple. The best architecture is like achems razor, i.e. "All things being equal, the most simple explanation is probably the correct explanation."
2) A language is just a tool -- a means to an end. There are many languages out there, and no single one is made to fit all problems. The language carries with it frameworks that are designed to solve specific problems. Find the right language for your problem domain.
3) Don't reinvent the wheel. If someone else has solved a problem or pattern, use it!
4) Admit what you don't know. Software is a broad, broad area. No one can know it all.
5) Keep up with your contacts. Your past associations are often your best resources for information, solutions, and (most importantly) friendship.
6) Understand the major design patterns. People often quote certain patterns in geeky conversations as if knowing the name of the pattern puts you in the "club". I have seen people list design patterns on their resumes. On close examination, they don't really understand them. Don't be one of those guys.
7) Learn the business domain. Just applying software principles to a problem may be enough to solve the problem, but it doesn't add value to your resume. Develop your expertise in the business itself. You will find that it will lead to more consulting situations over time than just knowing how to code.
8) Learn to code. Don't fake it.
9) RTFM. Documentation is done for a reason. You are better off spending some time learning the background information about your problem than just by looking at the code.
10) Contribute to the industry. You have an obligation to add to your craft! Develop your expertise and share it through publications, technical associations, instruction, etc.