Dick, Yes, I think only bug fixes should go into 1.0 RCs. Actually, I think once we get to RC stage, only really bad bugs (security, crashes) and their fixes should go into the RC. All other bugs should get pushed to a subsequent release.
Gianugo, Actually, it's not orthogonal at all. It's the original topic of the discussion ;-) And because of that, let's focus on topic #1 and forget that I mentioned #2. Though I think it's a valid concern, I recognize that if the mentors don't understand the concern, I must be missing something. Ethan On Tue, Mar 2, 2010 at 10:00 AM, Gianugo Rabellino <[email protected]> wrote: > On Tue, Mar 2, 2010 at 3:28 PM, Ethan Jewett <[email protected]> wrote: >> I only have two things to add here (assuming that this is the >> definition of a release within Apache): >> >> 1. My original concern: I think that nearly all the changes in JIRA >> that are assigned to Release-1.0-RC2 should be moved to something else >> called Release-1.1. We already agreed on a locked scope for release >> 1.0 and I don't think we should add anything to 1.0 release candidates >> aside from things we have agreed are blocking bugs. ESME-162 (mailto >> actions crash the server) is probably an example of something that >> should stay in Release-1.0-RC2. ESME-100 (finish Web UI) is an example >> of something that should *not* stay in Release-1.0-RC2. > > This is a valid concern, although orthogonal to the discussion here. > Still, yes, I would agree RCs should not contain any new features as > they might introduce bugs or regressions. > >> 2. Not to pick on our mentors, but this definition doesn't make any >> sense to me. It is aligned with the official Apache release definition >> at http://www.apache.org/dev/release.html#what but we've just moved >> the question from the definition of "release" to the definition of >> "the act of publishing it beyond the ESME group of developers (this >> mailing list)". If this is the definition of an Apache release, then >> the publicly accessible SVN repository is a release. I have a hard >> time believing that if I do an export from the ESME SVN repo and >> upload it to my people.apache.org page to facilitate testing that this >> constitutes a significantly different action from sending someone >> instructions on exporting the SVN repo themselves. > > As Richard pointed out, the real difference between "do an svn > checkout -r xxx" and "grab this tarball we just released" is consensus > coming from a community blessing by means of a vote. It's not peanuts, > it makes all the difference. > >> I suggest that we work with a narrower definition. Something like "a >> signed tarball published to http://www.apache.org/dist/incubator/esme/ >> and advertised on the public ESME website and/or the public mailing >> list is a release". > > You're more than welcome to argue your case, as no ASF procedure is > carved in stone, but know that you should make sure you place your > soapbox on front of the right audience - this is not the place to > discuss what the ASF, as a whole, considers a release to be - > gene...@incubator might be a better starting point. Until the current > definition stands, so does the current process. > > -- > Gianugo Rabellino > M: +44 779 5364 932 / +39 389 44 26 846 > Sourcesense - making sense of Open Source: http://www.sourcesense.com >
