> > > If issues are found during the discussion phase, you can just drop a > new package over the old package. > > -1 to this as this will cause all sort of confusion to exactly what > version an issue was in or what version people tested and this could more > RCs rounds not less and a lot more work for the release manager. There are > also issues with caching so that people may think they are testing the > latest version but are actually testing an older one. It's not a large > amount of effort to number them as rc0, rc1 etc etc but not call a vote > until the later rcs and there appears to be no outstanding issues. >
I'd go (and will go) one further, and use nightlies during this phase. These should always contain the latest HEAD from the release branch, therefor are "state of the art." All testers download the latest nightly before they start. I makes no sense to work on a week old version just because the release manager didn't have time to update the RC. This is one of the reasons we have automated builds upon every commit. EdB -- Ix Multimedia Software Jan Luykenstraat 27 3521 VB Utrecht T. 06-51952295 I. www.ixsoftware.nl
