Okay, I'm running under the assumption that the dev machine hits one of the NT boxen, while the test machine hits the other. Let me know if this is incorrect.

What's your $ENV{TDSVER} set to? With MSSQL 6.5, you should have both of the DECs set to '42'. I don't think this is related to your inconsistent DBI versions, since your test machine is actually higher versions of both Perl and DBI.

However, the key here may be the DBD version. I'm not 100% sure that FreeTDS has been tested with 0.23. The FreeTDS website lists 0.20, but it's been that way for a while. I'm also trusting that you followed instructions for building the libs.

You say that the code works on the dev machine, but not the test machine. Is it the exact same code? Finally, you may wish to subscribe to DBI Users mailing list. I've seen the problem you refer to, but aside from creating a second dbh, I've not solved it. My own problem lies in the fact that only about half of my subqueries produce this error, so I haven't been as perturbed by it.

ALL HAIL BRAK!!!