Justin Mclean <[email protected]> 于2019年12月4日周三 下午2:06写道：
> > Hi, > > I still slightly confused, but I'm obviously missing something. > > > They are doing tests before release check, so the binary tar should be > > packaged, source tagged. Then the team could test this tar(s), and make > > sure features are ready. > > Wouldn't they check that while doing the vote? Any connivance binary has > to 100% match the source code in the voted on source release. > They prefer to check the feature first maybe, I think the sequence is not required. If they want to check first, then they could do this. I think the key of our discussion here is more about the `announcement`. You are thinking they are doing release directly without IPMC vote. They are not. Do you have any place said, they should not use announcement mail, or just in English language perspective, it is not a good one? > Any connivance binary has to 100% match the source code in the voted on source release. They should for sure. But even they matched, the features could break. May be lack of autotests, or special env requires, whatever. I think how to test is the PPMC's choice. > > > Then they will use the same tar balls to do PPMC vote, then IPMC. > > It's highly unusual to vote on a binary release on its own. Apache > projects make and release source code not binaries. > I am not saying they are voting on binary. My point is, they are testing. > > It's very likely they will need to do another vote on the release, as in > it current form the source release is unlikely to pass an IPMC vote. (See > my previous email). > If you have found something wrong in their source release, it will be great to share it with PPMC. If they need another round of test and vote, then I think they will do. > > Thanks, > Justin >
