Should I bind my Heroku app to process.env.PORT or a fixed port for local development?

When developing a Heroku‑deployed application, I want the server to start successfully both on my local machine and when pushed to Heroku’s dynos.

The platform assigns a dynamic TCP port through the $PORT environment variable, while local runs often lack this variable, leading developers to either hardcode a port (e.g., 3000) or add a fallback.

What is the recommended approach for handling the missing $PORT in a local environment?

How can I verify that the fallback port works identically to the Heroku‑assigned port when testing with heroku local?

Are there any drawbacks to choosing a specific fallback port (such as 5000) that might interfere with other processes on the dyno?