How to manage a new feature in a separate branch? Feature itself being a size of project

I have a project with around 20 class files already. The client is asking for an experimental feature, which itself is as big as a project around 4-5 class files or even more. I don't want to connect the feature with the original project in the beginning.But later when I am sure, it works.

What is the right approach to handle such cases?

Should I create a new branch and remove all the 20 class files. And add new feature project there?

Or is there any different/better way to do this using git?

Or should I not use git for such case, and create the feature project independent of the original project?

Solution

For just a few additional files, a branch is a good approach, and will allow:

to isolate the development effort from the main development
to merge it back to the main branch eventually, if the experiment is deemed successful
to discard it if the experiment fails.

The other approach would be to fork the repository (GitHub), or simply duplicated it (regular git clone, but pushed to a new empty repository), in order to modify it directly.

is it conceptually right to "delete" those files, and start with a new project after deleting those files?

The main question is the end goal:

contribute back to the original project: using a branch is acceptable, even if the final merge completely change the codebase.
having a second separate project (while maintaining/releasing the first one in parallel): having a separate repository is simpler.