http://wikimedia.org/ is recognized. So if you have only a domain name, it *requires* a path.
This might be valid ([citation needed]) based on a strict interpretation of the RfC, but is problematic when users are inputting URLs we are validating/parsing.