There are sources published in hardy (at least one that I know of, "dom4j") that have no associated build record(s). This is, IMO, buggy behaviour.
As I understand it, queue-builder checks if a source had a build record in a previous series, and doesn't create a new one if it did. What it *should* be doing is checking if the source had a SUCCESSFUL build record in the previous series, and creating a new record if not.
This ensures both that transient failures get sorted on the new series, and that non-transient failures remain on the radar as "failures in the current release", both desirable outcomes.
From a history POV, I'd prefer to see a new record created for the new series, rather than the old record moved, since it would be nice for people like the SRU and security teams to be able to clearly see that $packagefoo had failed on $releasebar, and why.
I'm not sure if we will have time to address it during 1.2.1, but let's keep it in the radar until we know.