docker compose watch

Ever since I worked with docker, I was told to how bad bind mounts are. In the early days of docker, there used to be many blog posts about them and you could read several times a week how much of a bad person you're for using them. The docs are less opinionated about them nowadays, but they treated you as if you're literally a criminal. While in theory they can cause some troubles and destroy portability, practically speaking they are usually the preferred solution to persistent and accessible data. They are easy to use and provide a seamless integration into the host for many of us. But I also have to give in, they still cause me some headaches with permissions. I think it has to do with the user mapping, user/group ID, when both — the user on the host and the process in the container — need to have write access. This is usually the case while developing your application.

It looks like there are better ways nowadays. I just got to know about docker compose watch the other day:

The watch attribute automatically updates and previews your running Compose services as you edit and save your code. For many projects, this enables a hands-off development workflow once Compose is running, as services automatically update themselves when you save your work.

Once I start my containers with docker-compose up --watch, my local files directory is watched and every time I change or add something it will be synced to the container.

But it all depends on the action. I could also trigger a rebuild of the image, just restart the container or even exec a command every time things in my folder change.

Maybe it's time to stop using bind-mounts all together?