The patch I sent a few days ago doesn't work right, because when it's fetching something that it's never seen before, datum.getFetchTime() returns the *current* fetch time instead of the last fetch time. When it's fetching something that was fetched before, it returns the *last* fetch time. Obviously if you ask the web server for something that's modified since *right*now*, it isn't going to return anything.
This whole problem would go away if datum.getModifiedTime worked. When I dump the CrawlDatum out of the segment file, the modified time is definitely in there, but datum.getModifiedTime() seems to always return 0. If I find out why that's happening, I'll send a patch. -- http://www.linkedin.com/in/paultomblin