I think that's good advice, but the long term plan is to move away
completely from proto. Not because proto is bad, but I now think it's
now it's not the best fit for the problem (most of the objects don't
actually need to be mutable), and more people are familiar with S3/R5
which makes it a bit easier to get contributions.
Hadley
--
RStudio / Rice University
http://had.co.nz/