Bug solved!

Today, while working at home, I solved the bug that had me perplexed yesterday. I had the idea that perhaps it had something to do with the fact that it worked with Oracle 10g to 10g, but with Oracle 9i to 10g, it would hang. I made the hypothesis that perhaps 10g's import doesn't know when 9i's export file is complete. To test this hypothesis, I closed the named-pipe connected to the import process once all the data had transfered.

I compiled, set up the test, ran it, and voila! The import process completed, status was returned, and everything went swell.