21392 Rate this article:

Five Things To Do Before Writing Code


While I typically write about enterprise topics, I’m actually part of the Custom Solutions Group where we develop tools and applications for our customers. Many of our customers also develop their own tools and applications. It’s so tempting to just jump right in and start writing code, but a little work up front can save significant rework and errors later.

Here’s my list of the top five things to do before any code is written:

  1. Develop a Clear Understanding of the Problem
    Our customers often come to us with a solution without having fully defined the problem they need to solve. For example, the statement “We need to identify changes in land use over the last five years,” is really the statement of a solution. Make sure you are clear on the reasons you are undertaking the project. Is there a change in a law? Do you need some new data to support an experiment? Understanding the problem lets you explore a wider range of options in designing a solution.
  2. Identify and Understand the Users
    Make sure you understand the audience for the application as this will also impact your solution. How technical is the typical end user? What sort of environment will they be working in? How much training time will they have? If you can interview and consult the end users, you increase the chances of a successful project outcome. Also consider how many users the application will have and how long the application will be in use. For example, a data conversion tool that will be used on a monthly basis by a single user for 3 months has different quality and robustness requirements than a processing application that will be used by hundreds of people on a daily basis over the course of the next 3 years.
  3. Specify the Project Requirements
    This step is sometimes intimidating, but project requirements gathering can be as simple as interviewing the stakeholders and making a list of the features the solution needs to include. For example, “Must support GeoTIFF input and output”, "Window leveling capability", “Needs to run on Windows desktops”, and “Has to support Android Tablets” are all examples of requirements. Ideally, you will end up with a list of independent, unique items. Make sure to include any constraints, and don’t forget system and platform requirements. This includes the type of systems the application will run on as well as any specifics around graphics cards, memory and processor speed.
  4. Start the High Level Design
    While it’s premature at this stage to think about specific architecture and implementation details, you should start thinking about the user experience. This defines how a user will interact with your application. What will the user interface look like? How will the user move through the application? How will the different features interact with each other? Starting a storyboard of the application workflow can be very helpful.
  5. Select Appropriate Tools, Teammates and Stakeholders
    Define the tools you will be using on the project. This includes the programming environments and languages (e.g. IDL, Java Script, C++) you will use, as well as tools like source code control and defect tracking systems. Understand who will work on the project, including developers, technical writers and testers. If you think beyond the development team, you may also realize you need to include sales or marketing or even export compliance people in your project.

With some work at the beginning of the project to really define what you intend to develop, you’ll find that the rest of the project goes much more smoothly.