This bug is really annoying when you create your own cookiecutter package and need to generate the files again and again. You have to manual rm -rf the-dir before you can re-generate it. Is there anyone who can have a look?

Here was my attempt to fix the failing test: 19a8f2f.

It might not work on Windows, though.

@att14

Let step back to think about the requirement of this -f flag. I had a look for the document, and I noticed doc. It says Overwrite the contents of the output directory if it already exists.

Here is my understanding, if you decide to use -f flag, you give up any changes to the existing clone, and you re-start a new clone. So my initial fix was straightforward, deleting the existing clone.

The -f is not a frequently used option, because most of the time, you clone once, and you start to modify the clone. However, this flag is useful for template developers, they need to clone the template again and again to check the output content. During the testing/verification, they don't want to pick different project names, so the -f flag is very handy to let them use the same project name.

I am happy to hear your thought or anyone else from the committer group.

I don't disagree with you. I'm confused what the problem is. The change I made is to make your delete respect the -f option. You want to avoid deleting all the copy only directories if they haven't supplied -f.

I don't think there is a case the copy-on-dir exists and the overwrite_if_exists=False. Please check this block of code, if the cloned dir exists and the overwrite_if_exists=False, the program will raise an exception, but not copying/rendering anything. This means, if the code hits this line if os.path.isdir(outdir) and overwrite_if_exists:, the overwrite_if_exists must be True, and we don't need to add that check.

I also noticed there is any flag skip_if_file_exists, and I believe that is the one to control not overriding any modified content.

You are correct, but I see no reason to avoid being defensive since the check is in another function.

@zhongdai @att14 thank you for contribution. Rebased and merged.