Yea, we saw that too, but setting the no_compress option in the server didn't seem to work.

Thanks very much for all of your insightful help. That seems to have worked around the problem. I have e-mailed the maintainer of the package with your suggested fix, and (hopefully) our data parameters will never get bigger than the threshold (2^31 bytes) that we have set before the bug fix is released! :-)

Thanks again,
-nenbrian