Hello everyone! I’m evaluating different strategie...

# navigation-architecture-component

[Natasha Jordanovska]

Hello everyone! I’m evaluating different strategies for implementing conditional navigation in a Jetpack Compose application and would love to hear your thoughts on the best practices. Here are three approaches I’m considering:
1. Centralized Navigation Logic with LaunchedEffect: This method involves managing a nextScreen state in the ViewModel, observed in the UI layer through a LaunchedEffect. Navigation commands are triggered in response to state changes, ensuring all navigation logic is centralized.
2. Callbacks in ViewModel: In this approach, navigation is triggered directly from ViewModel functions based on API call results or other conditions, using callbacks. This method avoids passing navigation logic to the UI and aligns with practices demonstrated in the JetSurvey app from the official Jetpack Compose samples, which can be found here. The Composables also define their own callbacks for navigation, ensuring a clean separation of concerns.
3. Using LiveData with Observers in Composables: Adapting a traditional approach where the ViewModel updates LiveData based on conditions, and Composables observe these changes to handle navigation. Although this is more common in architectures involving fragments, it could be adapted for Composables as detailed in the Android developer guide here.
I’m particularly interested in scalability, testability, and ease of maintenance. Which approach do you prefer for conditional navigation in your projects, especially when using Composables, and why?