Don’t code against the UI

In modern application development, one common pitfall is coding directly against the User Interface (UI). This approach results in tight coupling between the logic of the application and the visual representation of that logic. But why is this a problem, and how should one approach the challenge of managing state in a user interface?

What Does “Coding Against the UI” Mean?

“Coding against the UI” means directly interacting with UI elements to read and update values. For instance, if you have a form with several input fields, instead of reading and writing data directly from and to these fields, you should have a representation of this data behind the scenes. The issues with directly coding against the UI are:

  1. Tight Coupling: When the UI changes, the logic tied to those UI elements must also change. This is not maintainable in the long run.
  2. Lack of Flexibility: The UI is not always consistent. Consider responsive designs where the UI might look different on a phone versus a tablet or desktop.
  3. Testing Challenges: Directly tied logic makes it difficult to unit test without involving the UI.

The Better Approach: Decoupling with a ViewModel

To address the above concerns, introduce a layer behind the UI: the ViewModel. This layer represents the state of the UI and acts as a cache for data. The benefits are:

  1. Separation of Concerns: The UI solely becomes responsible for displaying data, while the ViewModel takes care of the logic and state.
  2. Flexibility: As the ViewModel is decoupled from the UI, you can easily switch or modify the UI without affecting the underlying logic. Think of scenarios like changing the UI based on device orientation.
  3. Testability: With the logic separated from the UI, unit testing becomes more straightforward.

How Does This Work in Practice?

Instead of directly reading from or writing to UI components, use a binding engine. This engine keeps the UI and ViewModel in sync. When a user interacts with the UI:

  1. The binding engine updates the ViewModel.
  2. Any logic that needs to run based on this change can be executed in the ViewModel.
  3. The ViewModel updates, and through the binding engine, the UI reflects these changes.

This cycle ensures that the UI remains a reflection of the ViewModel, and not the other way around.

Understanding Modern UI Architectures: From MVC to MV*

Modern software development emphasizes the separation of concerns, ensuring that applications are modular, maintainable, and scalable. A key player in this evolution has been the family of MV* patterns: MVC, MVP, MVVM, and more. Let’s dive into these patterns and see how they guide the development of robust software solutions.

1. The Essence of Not Coding Against the UI

To “not code against the UI” means to avoid embedding business logic directly within the user interface code. By keeping business logic separate, developers can:

  • Modify the UI without affecting core functionality.
  • Test core functionality without complex UI interactions.
  • Reuse logic across multiple interfaces or platforms.

2. MVC (Model-View-Controller)

MVC divides an application into three interconnected components:

  • Model: Manages the data and business rules of the application.
  • View: Displays the data to the user.
  • Controller: Interacts between the Model and the View.

For example, in a weather app, the Model could fetch weather data, the View displays it, and the Controller handles user requests like changing the city.

3. MVVM (Model-View-ViewModel)

A variation of MVC, MVVM is mainly used with frameworks like WPF and Angular.

  • Model: Represents the data.
  • View: Displays the data to the user.
  • ViewModel: Acts as a bridge, exposing the data from the Model in a way that’s easy to manage and present in the View.

For instance, in a messaging app, the Model handles message storage, the ViewModel could handle formatting and grouping messages, and the View displays them.

4. The Role of the Binding Engine

In MVVM and similar patterns, a binding engine is vital. It ensures synchronization between the data model and the UI. Here are its key roles:

  • Automating UI updates based on data changes.
  • Reducing repetitive code to update UI elements.
  • Ensuring UI consistently represents the current data state.
  • Supporting two-way data binding for responsive applications.

5. Flexibility of Presentation

By separating business logic from UI, developers can create multiple interfaces for the same functionality. You can have a traditional web-based UI in HTML and a separate 3D graphical environment for VR, both powered by the same underlying logic.

6. The MV* Family

MV* takes inspiration from both MVC and MVVM, combining their strengths to offer a more flexible architecture. Here’s a breakdown of its components:

  1. View: This is the user interface (UI) component. It displays the data and receives user input. In MV*, the view is a passive interface that displays data provided by the ViewModel and sends user commands to the ViewModel or Controller. It doesn’t have any logic but merely presents the data.
  2. ViewModel: Acts as a bridge between the View and Controller. It deals with UI-specific logic, ensuring that the data is in the correct format for display. For instance, if a date needs to be shown in a particular format or a number needs rounding for display, the ViewModel takes care of it. The ViewModel is also responsible for reacting to user inputs and updating the View accordingly.
  3. Controller: While the ViewModel handles UI logic, the Controller is concerned with business logic. It processes user commands, interacts with the Model to fetch or persist data, and updates the ViewModel. By keeping business logic out of the ViewModel, MV* ensures a clean separation of concerns, making the system more maintainable and scalable.
  4. Model: Represents the data and the business rules of the application. It communicates to the database and updates the Controller about the changes in data. In MV*, the Model remains largely untouched from its role in MVC or MVVM, serving as the single source of truth for the application’s data.

Benefits of MV*:

  • Flexibility: MV* isn’t rigid. Depending on the application’s needs, developers can adjust the responsibilities of the ViewModel and Controller, ensuring the best fit for their particular scenario.
  • Separation of Concerns: By clearly distinguishing UI logic (ViewModel) from business logic (Controller), MV* promotes modular code. This separation makes the codebase more maintainable and easier to test.
  • Reusability: Components, especially the Model and Controller, can be reused across different parts of an application or even across different applications.


MV* offers a nuanced approach to application architecture, allowing developers to harness the strengths of both MVC and MVVM. While it may not be the definitive solution for all scenarios, its flexibility and clear separation of concerns make it an attractive option for many complex applications.