Currently, this package does not protect strings when they are written to score files. That means evaluate.py cannot do its job in score files with spaces on identities or labels such as the one described in bob.measure#36 (moved).

Since bob.measure.load uses csv.read to load score files (please confirm), the robust way to handle this could be like this:

1. Fix bob.measure#36 (moved)
2. In this package, protect strings dumped during scoring in case they have spaces, using a " (double-quotes) delimiter