I've started trying stuff within my hosting machine so I'd more confidently build an image later. It didn't pay off. Just type inside the Dockerfile instead of the terminal (shell).
Nevermind the state of the art when you're starting off:
- Maximize the number of layers when you're playing around with stuff so you
benefit the most of cached layers -- no need to run each command within a
giant
RUN
directive
And then
- Optimize (e.g.: wrap up everything within a single
RUN
) when you finished tweaking
First lessons could probably be generalized to anything else: