For CVS administrators, that would be nice to run statcvs on the cvs server and compute the stats directly from the RCS files in the repository instead of requiring a working copy of the project.
