Gimmie The Code!

  • Comments 0

It seems like a question that has an obvious answer: How should you show code listings in guidance documents? I'm not talking about the C#/VB/other language debate, or whether you orient it in landscape mode to avoid breaking long lines. No, I'm talking about the really important topics such as what color the text is, how big the tabs are, and where you put the accompanying description.

We're told that developers increasingly demand code rather than explanation when they search for help, though I'm guessing they also need some description of what the code does, the places it works best, and how to modify it for their own unique requirements. However, copy and paste does seem to be a staple programming technique for many, and is certainly valid as long as you understand what's going on and can verify its suitability. I've actually seen extracts of code I wrote as samples for ASP 2.0 when it first appeared (complete with my code comments) in applications that I was asked to review.

But here in p&p a lot of what we create is architectural guidance and design documentation designed to help you think about the problem, be aware of the issues and considerations, and apply best practice principles. As well as suggesting how you can implement efficient and secure code to achieve your design aims with minimal waste of time and cost. "Proven practices for predictable results", as it says on the p&p tin.

But even design guidance needs to demonstrate how stuff works, so we generally have some developers in the team who create code samples or entire reference implementation (RI) applications. These are, of course, incredibly clever people who don't take kindly to us lowly documentation engineers telling them how to set up their environment, or that the code comments really should be sentences that make sense and have as few spelling mistakes as possible.

In addition, Visual Studio has a really amazing built-in capability that we've so far failed to replicate in printed books. It can scroll sideways. These esteemed developers often prefer to have four or more space character wide tabs to make it easy to read the code on screen (the Visual Studio default is four). By the time you are inside a couple of if statements, a try/catch, and a lambda statement, you're off the page in a book. Two spaces is plenty in a printed document (where we have to replace the tabs with spaces anyway), but I've never yet persuaded a developer to change the settings.

And now Microsoft mandates that we have to use the same colors for the text in listings as it appears in Visual Studio (I guess to make it look more realistic, or at least more familiar). The old trick of copying the code into Notepad, editing it, and then pasting it into the document no longer works. But copying it directly from the Visual Studio editor into a document is painful because it insists on setting the font, style, margins, and other stuff when I just want to copy the colors. Yet if I do Paste | Special | Unformatted Text in Word, I lose the colors.

And then, when I finally get the code into the document, I need to describe how it works. Do I dump the entire code for a class into a single code section and describe it at the start, or at the end? If the code is a hundred lines or more (not unusual), the reader will find it cumbersome to relate parts of what is likely to be a long descriptive section to the actual code listing. I can break the class down into separate methods and describe each one separately, but often these listings are so long that I have the same problem.

And, of course, explaining how the methods relate to each other often means including an abridged version of some parts of the class or one of its methods, showing how it calls other methods of the class. But do I list these methods first and reference back to them, or explain the flow of execution first with the abridged listing and then show the methods that get called?

Typically I end up splitting the code into chunks of 30 lines or less (about half a printed page) and insert text to introduce the code before the chunk and text to describe how it works after the chunk. Something like:

The GoDoIt method shown in the code listing above calls the DoThisBit method to carry out the first operation in the workflow. The DoThisBit method takes a parameter named thisAction that specifies the Task instance to execute, as shown in the following code listing.

[CODE LISTING HERE]

The DoThisBit method first checks that the task is valid and then creates an instance of a ThisBitFactory, which it uses to obtain an instance of the BitHelper class... and so on.

After going backwards and forwards swapping code and text, breaking it up into ever smaller chunks, and trying to figure out what the code actually does it's just a matter of editing the code comments so that they make sense, breaking the lines in the correct places because they inevitably are longer than the page width, and then persuading the developer to update the code project files to match (or doing that myself just to annoy them).

Sometimes I think that putting code listings into a document takes longer than actually writing the code, though I've never yet managed to convince our developers of that. But I've been doing it for nigh on twenty years now, so I probably should be starting to get the hang of it...

Leave a Comment
  • Please add 1 and 8 and type the answer here:
  • Post

Gimmie The Code!