Good Examples of and Practices in Code Documentation

Question

I noted in this question:

What is the best way to learn technical writing?

...that a rated method to learn how to write technical documents is to imbibe good examples.

Can you recommend any examples of good code/ database documentation that have proven useful?

This was also a really useful thread in terms of documenting code: How can I write better code-based reference documentation for programming interfaces?

I am documenting a large number of programs so that developers can rewrite and maintain the code and database. The documentation will be directed at developers who will be maintaining and re-writing the code, rather than end-users. Any advice, dos/ don'ts, words to the wise much appreciated.

Answer 1

Introduction

The Biggest reason documentation is written is to help developers learn about the software system and give them a reference to the tools they are using. This is a broad question and I must admit most of the tips I will give will be my opinions and things I've found helpful. Below are some guidelines and design aspects you can use to help improve you documentation.

Know Your Audience

This may sound like a silly thing but it is very important. Knowing your audience helps you decide how much detail you need and what information you need to provide. For example, You don't want to present all this unnecessary low level details about your system, and techno jargon to someone who just wants to fix little bugs in your system or test your programs. You also want to accommodate for the advance developers you want to really change your system, they don't want to many abstractions and not enough detail about a system.

Django is an example that does a good job identifying their audience and accommodating for their different targets.

Make It Easy to Find Information

When developers get stuck on a problem or want a quick look at what arguments some function takes or what it does, they don't want to have to look through thousands of links just to find what they are looking for. Just like you might not want to read all of this answer if I didn't have headers. Creating an easy navigation system for your documentation makes it clearer. I would suggest a contents table or a searching function. Also creating a structure for you documentation help readers find what they are looking for better.

Android does a terrific job with this. Dividing their content into Design, Develop, and Distribute gives developers and idea of where to find the information they are looking for. Their search function makes finding information easier.

Spruce It Up

Give you audience a choice on how they want to receive the info. Not everyone wants to read long lines of text. Some people might want to watch videos, others might want to look at an infographic. Even if you you just want to use text then put your documentation in the form of a story, or make it more personal.

Qt does a good job providing different mediums of documentation. Ooc programming language does a good job sprucing up their tutorial by making it into a story.

Closing

Writing documentation is boring and difficult but it can be very helpful to your software. Just always remember who you are writing for and why you are writing. You can use any guidelines you want. Hope you found this helpful and I would welcome any criticism.

Answer 2

In test-driven development the emphasis is on writing tests that clarify what code should do, rather than extensive documentation. Future maintainers of the code can then ensure that the code continues to exhibit the behavior it should by virtue of the tests.

The way tests are named form a type of documentation of the proper behavior of the code. Here is an example from the Ruby world:

describe "REST interface" do
  describe "create" do
    context "with valid parameters"
      it "should create an object in the database" do
        # test logic goes here
      end
      it "should return HTTP 200" do
        # test logic goes here
      end
    end

    context "with invalid parameters" do
      it "should not change the DB" do
        # test logic here
      end
      it "should return HTTP 400" do
        # test logic here"
    end
  end
end

The block names are chained together, so that you can read them in English and they make sense: REST interface create with valid parameters should create an object in the database.