Hi; I'm not a committer yet but I want to share my thoughts from a perspective of a user. I've been using SolrCloud since 4.1.0 version of it. I've read nearly all e-mails and I follow mail list too. Solr project has a great development cycle and has a frequent release cycle. In fact, if you compare it with some other Apache Projects it is has really nice commit rates. I've prepared a chart that explains the release cycle of Solr since 4.0 and attached it to this e-mail to make everything clear.
When you check the chart that I prepared you will see that Solr has followed that release cycle(for 4.x releases): If needed it has always had bugfix releases. So except for 4.0, 4.1.0 and 4.4.0 it had bug fix-releases (I do not include 4.7). However bug-fix releases are applied once for each main release. I mean there is no 4.3.2 after 4.3.1 or 4.6.2 after 4.6.1 When you use a project as like Solr you should catch up the current release or current stable release (as like a bugfix release). I think question should be that. If somebody finds a bug at a bugfix release what will happen? Will be a 4.x.2 release or it will be resolved with 4.x+1.2? I also think that solution can be that: maintaining 4.x.1 and applying changes to both for 4.x+1.0 and 4.x.2 So if anybody wants to use new features (of course with recently bug fixes) and accept the risk of new features user can use 4.x+1.0 otherwise a more stable version: 4.x.2 This causes a new question. What will be the limit for "*y*" at 4.x.y? As a perspective of a user who uses Solr and tests and checks its all versions my thought is that: 2 (or 3) may be enough for that. Long term support is a good idea (if you accept value of "*y*" as 2 or 3 it will be 4-6 months). Solr is developing so fast and it has nearly good features that users really need it. "If maintenance is not a problem" to apply bug-fixes to a release of 4.x.2 and 4.x+1.0 having a "*y*" vale that is greater than "1" may be a solution. If we just say that: "this release will be long term supported" -I think that- people will want to use new releases after a time later because of the new features nowadays. On the other hand if we release more than 1 bug-fix releases and if people do not need new features they will have a more stable version of their current version and will be able to use it. Thanks; Furkan KAMACI 2014-03-12 18:34 GMT+02:00 Mark Miller <markrmil...@gmail.com>: > +1 to the idea, I love bug fix releases (which is why I volunteered to do > the last couple). > > The main limiting factor is a volunteer to do it. Users requesting a > specific bug fix relese is probably a good way to prompt volunteers though. > > -- > Mark Miller > about.me/markrmiller > > On March 12, 2014 at 9:14:50 AM, Doug Turnbull ( > dturnb...@opensourceconnections.com) wrote: > > Hello Solr community, > > We have been using Solr to great effect at OpenSource Connections. > Occasionally though, we'll hit a bug in say 4.5.1, that gets fixed in > 4.6.0. Unfortunately, as 4.6.0 is a release sporting several new features, > there's invariably new bugs that get introduced. So while my bug in 4.5.1 > is fixed, a new bug related to new features in 4.6.0 means 4.6.0 might be a > showstopper. > > This is more a question for the PMC than anything (with comments from > others welcome). Would it be possible to do more minor bug-fix releases? I > realize this could be a burden, so maybe it would be good to pick a > version and decide this will be a "long term support" release. We will > backport bug fixes and do several additional bug-fix releases for 4-6 > months? Then we'd pick another version to be a "long term support" release? > > This would help with the overall stability of Solr and help in the decision > about how/when to upgrade Solr. > > Cheers, > -- > Doug Turnbull > Search & Big Data Architect > OpenSource Connections <http://o19s.com> >
