I'm just using DBD::mysql, which should be the best thing for the job. Further, the program doesn't even make it into the loop, as the 'print' diagnostic I have in the code doesn't report back. Instead, a HUGE amount of data is sent from the MySQL server over TCP/IP to the Perl process, which happily gobbles it up into RAM.

ichimunki's idea of the brackets for the my declarations was interesting, but produced no effect. I even removed the error checking from $s->execute() and there was no change.

The only solution (read: hack) that I came up with is to limit the query to managable chunks:       SELECT * FROM table LIMIT ?,? Then iterate through the entire table in LIMITed chunks. This is not the fix I was hoping for.