Currently, to drop labels along an existing dimension, you need to write something like: ds.drop(['a', 'b'], dim='x).

It would be nice if keyword arguments were supported, e.g., ds.drop(x=['a', 'b']). This would make drop() more symmetric with sel().