FWIW, I don't think this bug has any serious conflict with my earlier report (bug 120052) which probably led to this state. I apologise for not considering binary package handling properly at that point. If Soyuz does what I said in bug 120052 for source packages and does what Adam says for binary packages then that will be just fine.
The important difference is that only contrib/non-free->multiverse component mappings should apply for binary packages, not the main->universe component mapping (since main is an Ubuntu component name as well as a Debian component name). I would suggest that a reasonable implementation strategy would be for the code handling new binary packages to check whether the component is valid in the target distribution before applying component overrides.
FWIW, I don't think this bug has any serious conflict with my earlier report (bug 120052) which probably led to this state. I apologise for not considering binary package handling properly at that point. If Soyuz does what I said in bug 120052 for source packages and does what Adam says for binary packages then that will be just fine.
The important difference is that only contrib/ non-free- >multiverse component mappings should apply for binary packages, not the main->universe component mapping (since main is an Ubuntu component name as well as a Debian component name). I would suggest that a reasonable implementation strategy would be for the code handling new binary packages to check whether the component is valid in the target distribution before applying component overrides.