Power Up Your Code Documentation with Fusedoc
Fusedoc is a standardized method for documenting the code to be written for a ColdFusion template. Created by Hal Helms, Fusedoc was originally designed to be used within applications written under the Fusebox application framework. However, Fusedoc has a great deal to offer, regardless of whether you use Fusebox in its entirety.
The Essence of Fusedoc
We have tried to teach the concepts and importance of Fusedocs in a number of wayscute games, outright preaching, plentiful examplesbut the best we have come up with is a simple metaphor. Sitting in our office one day, we noticed that things were getting interesting on the construction site next door. The foundation hole was getting bigger every day, and rumors made the rounds of the office about what was being built. We thought it would be great to see the blueprints for whatever was going to end up there. Then, the lightbulb went offblueprints and parts lists are the Fusedocs of the construction world!
Actually, it's more the case that Fusedocs are the blueprints and parts lists of the Fusebox world, but we won't worry about splitting hairs. We will worry about what Fusedocs are, how to create them, and the enormous power they offer to our development projects. For the uninformed, Fusebox is a standardized system, at the heart of which is simplicity: a framework consisting of a set of helper files and organizational principles, and a methodology consisting of a set of best practices for managing web projects. Used by application designers, developers, and architects, the system addresses development problems, such as unmanageable complexity, wasteful redundancy of effort, time-consuming code maintenance, and slow development speed. More information on Fusebox can be found in our book Fusebox: Developing ColdFusion Applications, from New Riders (2002).
Fusedocs bring the power of communication and documentation to Fusebox. Through Fusedocs' use, applications can carry a wealth of information that is available both to humans who are working on the project and to programs that are designed to read the standard Fusedoc format.
What Is Fusedoc?
Fusedoc is a documentation standard developed by Hal Helms and patterned originally after the JavaDoc concept. Since its inception, Fusedoc has grown into the current Fusedoc 2.0 standard, which is an XML vocabulary. Fusedocs appear at the top of every fuse file, providing detailed information about the fuse and its input-output requirements.
For architects, Fusedocs represent a blueprint concept, allowing careful planning of every fuse in an application. For coders, who might not get to see the entire application, Fusedocs represent more of a builder's parts list, providing detailed instructions on what the fuse needs to do. Note, however, that Fusedocs do not tell the coder how the fuse should work, only what it should do.
Some developers feel that Fusedoc represents too much overhead work. In reality, well-constructed Fusedocs represent the potential for decreased workload overall, and a much more robust system. This is because Fusedocs are part of the planning process for a Fusebox application. Their power is based on the architect's initial creation and everyone on the project's consistent understanding that the Fusedocs are the technical definition of the application. Consequently, the practice of using Fusedocs is structured and flows through every aspect of the development process.
The Tao of Fusedoc
When we use Fusedocs, we have expectations about how they will be used. Consistency is the key to successful Fusedocs. Fusedoc is a standard for embedded program documentation. That is, every fuse in a Fusebox application should have a Fusedoc embedded at the top of the file. This ensures the ability to easily examine the documentation for any fuse. Listing 1 shows a sample Fusedoc as it might appear in a typical fuse.
Listing 1Typical Fusedoc
<!--- <fusedoc fuse="act_Login.cfm" language="ColdFusion" Ëspecification="2.0"> <responsibilities> I validate a user's login information. </responsibilities> <properties> <property name="Date" value="01 Jan 02" comments="Sample for Ëbook"> </properties> <note> Notes can be used to capture information that doesn't have a Ëspecific Fusedoc element. </note> <io> <in> <string name="XFA.onSuccess" optional="No" comments="Use if Ëprocess succeeds"> <string name="XFA.onFailure" optional="No" comments="Use if Ëprocess fails"> <string name="userID" optional="No"> <string name="password" optional="No" comments="Hash() this Ëstring before comparing to password in database"> </in> <out> <string name="userID" scope="client" oncondition="User is Ëvalid"> <string name="firstName" scope="client" oncondition="User is Ëvalid"> <string name="email" scope="client" oncondition="User is valid"> </out> </io> </fusedoc> --->
Fusedocs are written before the application's code is written. We will get into this in much detail shortly; the important idea is that the emphasis on Fusedoc comes during the architectural design stage of the development project, as opposed to the more traditional approach of documenting code as it is written.
The process of writing Fusedocs is an analytic and creative one. The objective is to create a blueprint for the application that coders can then pick up and work from. In truth, Fusedocs are more like a parts list that a contractor gives a builder. Fusedocs tell builders exactly what parts will be needed to successfully complete the job in question. That is what Fusedocs dodetail the variables and data that are available to the coder, along with some explanation of what is expected from the fuse in question. Armed with this information, the coder is then able to build the fuse without needing to consult with outside resources.
If the coder is supposed to be able to write the fuse without consulting with outside resources, it is clear that a Fusedoc needs to be explicit about what the architect expects from the fuse. Fusedocs are about what the fuse is to do, not about how to do it. This is a fine line to walk, but it is important.
For example, the Fusedoc in Listing 1 tells the fusecoder to validate the user according to the userID and password. The fusecoder might choose to run a SQL query against the database that asks for records matching the userID and password and then check the output for records. Another possibility is to loop over a recordset of userIDs and passwords, looking for a match. The approach that works best is up to the fusecoder.
The creation of good Fusedocs is a skill learned only through the experience of having others write code according to your Fusedocs. In this regard, writing Fusedocs can be considered an art.
The Art of Writing Good Fusedocs
Although software development is generally recognized as a technical discipline, successfully writing good Fusedocs depends on experience, on the refinement that comes only through trial and error.
We all make assumptions about the systems we are building. For that reason, a sort of motto for writing good Fusedocs has come to be "When in doubt, put it in." A word of caution, though; this does not mean that we should include every variable imaginable in every fuse's Fusedoc. In fact, we want to do exactly the opposite. We want to include everything the coder needs to write the fuse, and nothing more. We state this as the golden rule of Fusedocs:
Include everything a coder needs to write the fuse, and nothing more.
Everything the Coder Needs
We have habits about the way we work that we might not realize we have. For instance, we sometimes use a variable named request.dsn to store the name of the ODBC datasource that our system is using. This makes it easy to transport a query fuse from one application to another.
The use of these variables makes sense, but it can cause a problem when we start writing Fusedocs. The fact that we use these variables as part of every query we write means that they become a sort of background noise in our coding environment. They are always there, and we can easily forget about their importance. Forgetting to include these variables in a Fusedoc that we write for someone in-house isn't a huge problem. The in-house coder knows our local conventions and will probably write the query after that fashion, so no harm is done. On the other hand, if we outsource the fuse for coding, the coder has no idea about the common use of those two variables in our shop, and he cannot successfully complete the fuse without them. Again, a Fusedoc must include everything the coder needs to successfully write the fuse.
To be effective Fusebox architects, we need to be familiar with the elements that compose the Fusedoc vocabulary. These elements can be found in the Fusedoc document type definition (DTD), available at www.fusebox.org.