Is there a way to ignore directories?

[TheThirdOne]

If not that should be added; otherwise it should be documented.

[AaronO]

All files listed in either .gitignore or .ignore will be ignored.

We could also add .bookignore to that list. That would specifically be for GitBooks. What do you think ?

[TheThirdOne]

I think .bookignore should be added simply to have a GitBook specific file, but I also think a command line option for ignoring certain files would be helpful if that isn't too hard to implement.

[AaronO]

I don't want to make the command line tool too cumbersome to use.

In my opinion a good convention is far superior to configuration.

I would rather have people use a .bookignore than remember that they have to add --ignore=folder1/,some_other_file etc ... (I also think that the later is uglier and more error prone, thus less desirable).

Do you have a strong and unique use case where a command line flag that would be better that a .bookignore file ?

[AaronO]

Basically all GitBooks should have the same build commands. With a good convention it's easier for new comers to pick up a new book, since they're already familiar with GtiBooks conventions and standards and don't need to learn the project's specific standards.

We need to do a good job at keeping and enforcing these conventions, because that's what makes GitBook "beautiful" and "elegant" ;)

P.S: I just wanted to help you understand where I'm coming from on this matter.

[TheThirdOne]

I definitely see where you are coming from, and agree a .bookignore should be the preferred method for projects, but the dependency of a file shouldn't be a requirement. I don't see the point in restricting the command line interface for the purpose of simply maintaining good convention.

[AaronO]

The command line isn't restricted.

In this specific case, I believe that in most scenarios people would rather persist the ignore options than have to remember to include them at each build/serve/whatever ...

Not persisting those options, is also a bad practice. Because then collaborators can't contribute to your book without knowing the specific folders you ignore when running the gitbook command.

I can't think of a single compelling use case, where providing that info as a command line flag is superior (in terms of simplicity and general usefulness), that maintaining a .bookignore file.

Users are already used to .gitignore, .npmignore and others. So I think they would rather opt for the .bookignore option. And thus an --ignore flag would go underused on top of leaving the door open to bad practices.

I don't think we should add any features without a compelling use case behind them.

I'm sorry for being strict on this matter. But I really would like to avoid feature creep, and instead preserve strong conventions and best practices.