Here's my (non-binding) -1 for a 2.9 beta. Before Lucene started using an X.Y.Z release naming process (v1.4 or thereabouts), releases generally had multiple release candidates. This left Lucene in quasi-released limbo for long periods of time. My take on the switch to an X.Y.Z release process was that it allowed people to accept releases as true releases. With the possibility of X.Y.(Z+1), we can have fast turnaround on discovered issues by putting out bug-fix releases. IMHO, what a new release needs is lots of eyeballs, and calling it a beta doesn't help.
Mike M.'s suggestion of a longish beta period for 2.9 could be interpreted/implemented in (at least) three ways: 1. Traditional (in other projects' traditions, anyway) beta release named "2.9-beta". (This is what I'm -1 on.) 2. Holding off on a 3.0 release, maybe for 2-3 months, until we're sure that 2.9 is stable; maybe release 2.9.1+, if necessary. (This gets my +1.) 3. Don't hold off on a 3.0 release, but allow for the possibility of parallel development: 2.9.X is maintained on a branch, and may have release(s) interspersed with the trunk (3.X) releases. (Lucene has never done this before - parallel branch release maintenance would be a new thing.) Steve > -----Original Message----- > From: Jason Rutherglen [mailto:jason.rutherg...@gmail.com] > Sent: Tuesday, August 11, 2009 12:51 PM > To: java-dev@lucene.apache.org > Subject: Re: Beta (was Re: who clears attributes?) > > I thought 2.9 was on track to be released shortly, then the new > query parser and the further tokenizer changes (which were > apparently necessary) went in. It seems like we're putting too > many additions into this release at the last minute (which I > thought was targeted to be the end of the month?). > > When we say beta testing we're talking about which > functionality? Many of the additions/changes to 2.9 have been in > for months or are very simple? > > I like all the new functionality however it seems like we could > easily be delayed 1-N months which seems a bit long for a Lucene > release cycle. If 3.0 was going to be the switch over to new > features and deprecating old ones, the tokenizers and query > parser probably should have been released then? > > On Tue, Aug 11, 2009 at 8:39 AM, DM Smith<dmsmith...@gmail.com> wrote: > > On 08/11/2009 08:22 AM, Michael McCandless wrote: > >> > >> I do still think a longish 2.9 beta is warranted, if we can succeed > in > >> getting users outside the dev group to kick the tires and uncover > >> stuff. > >> > > > > I think a beta would be a great idea. Not sure it needs to be > "longish." > > > > Having not looked at it, I'm a bit leery about the token/filter > changes and > > new query parser. To me, a user and very minor contributor, I don't > see much > > value in it, see potential cost to me, and fear a performance hit > (most of > > my docs are a sentence in length). Because it has been changing on a > regular > > basis, I have not been motivated to look at it or play with it. So > hopefully > > my trepidation is unfounded. > > > > But, with the long release cycles of Lucene, the future value seems > to far > > away. It might be good to prepare the user community for the change. > > > > If, as another note suggested, 2.9 is sometime in the next 30 days, > when > > would it start. What needs to be done before a beta? > > > > How about putting out an alpha now? It would have the normal alpha > caveats. > > > > I think starting on the actual path to release may cause it to happen > sooner > > rather than later. I am eager for a release. > > > > > > -- DM --------------------------------------------------------------------- To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org For additional commands, e-mail: java-dev-h...@lucene.apache.org
