> Hm, I took a look on both thread about patch and it seems to me now it's > overcomplicated. With recently committed enhancements of pgbench (\if, > \when) it becomes close to impossible to retry transation in case of > failure. So, initial approach just to rollback such transaction looks > more attractive.
I think that the best approach for now is simply to reset (command zero, random generator) and start over the whole script, without attempting to be more intelligent. The limitations should be clearly documented (one transaction per script), though. That would be a significant enhancement already.