To give some context. Before the project can release code under the protection of the foundation, there must be three +1 votes by members of the PMC.
The legal requirement for invoking the protection of the foundation says nothing about how many -1 votes there are. In fact we could have a release with 50 x -1 from the community, 20 x -1 from committers, 10 x -1 from the PMC and 3 x +1 by the PMC and the legal requirement has still been met. Now obviously I think it would be bad form for the release manager to publish a release with 80 x -1 votes and only 3 x +1 votes, but the foundation policy says you cannot veto releases and unless the Maven project decides to adopt local conventions... it all becomes the decision of the release manager to go ahead with the release or not. Some things people should also think about: * Should we have a project mandate that a release manager can only go ahead with the release when there is a majority of votes cast in favour of releasing? (my view is a recommendation is sufficient, there may be cases where the release manager has a valid reason to go ahead with the release even if there is a majority of -1 votes) * Who should be valid votes for the above? (my view is that all committers and contributors to the release should have their vote counted) On 29 October 2015 at 06:58, Hervé Boutemy <[email protected]> wrote: > Dear Apache Maven Community, > > Recently, in a discussion within the PMC, we discovered a complete > misconception about release votes: a lot of people feel that voting on > releases is not useful if you are not in the PMC. That's wrong! > > After digging into the issue, we found that a few years ago, when we improved > our legal requirements on releases (by ensuring vote on source > distribution and not only convenience binary and that we had a minimum of 3 > PMC votes), we introduced a misleading wording in our voting template about > this "3 PMC minimum votes" requirement: we called them "binding", then counted > "binding vs non-binding" votes. Yes, these "non-binding" votes are not binding > regarding legal requirements, but they are the most useful feedback we are > looking for when we vote on a release = does it work in real users > environment? > > In conclusion, we decided: > > 1. to change our wording and remove this "binding vs non-binding" count: we > still need to find a new wording > > 2. to explain the issue to the whole community to get feedback > > You already saw some experiments around this in recent votes. > > We need your feedback on this change, we need your feedback on release votes: > in short we need your involvement. Don't hesitate to tell us if we missed > something, or to share ideas. > > We hope this will improve future evolutions of our beloved Apache Maven > project with the help of the whole community. > > Regards, > > Hervé > Apache Maven PMC Chair > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
