>>> On 6/19/2008 at 2:04 AM, in message <[EMAIL PROTECTED]>, Carlo Marcelo Arenas Belon <[EMAIL PROTECTED]> wrote: > On Tue, Jun 17, 2008 at 09:33:26AM -0600, Brad Nicholes wrote: > >> It's time to get Ganglia 3.1 out the door. > > Brad, by the time you made this statement : > > * AIX users had no working build > * Fedora Linux ppc64 users wouldn't be able to build a package > * OpenSuSE 10.3 users are still figuring out where to get a working > libconfuse > * MacOS X 10.5 users have no working build, and users of older versions have > a broken libconfuse dependency and most likely a broken build > * HPUX users have no working build > * OpenBSD amd64 users will get their modules installed in an odd directory > that doesn't exist otherwise in their systems > * DragonFlyBSD users had never had the luxury of having a working snapshot > to use for testing. > * Solaris users trying to reuse the libapr that comes with apache will have > a broken build because of an internal header conflict. > * CentOS 4 users are getting used to the idea of using some packages they > had > been told are not safe, as dependencies. > * Windows users had only a static build to use. > > what makes you think ganglia 3.1 is ready to get out of the door? >
I disagree. Release early, release often. There will always be issues to resolve. How long do we want to wait for somebody to step up, identify and fix all of the issues. These are all issues that can or could be resolved in future minor releases. If we waited for all issues to be resolved on all platform, Ganglia 3.1 would never be released. The problem is that we don't have active developers who are responsible for every platform and those of us who are active, are doing the best we can on the platforms we support. Until we do get community members to step up and fix issues on platforms that are important to them, it isn't fair to hold the rest of the supported platforms hostage. The bottom line is that Ganglia 3.1 has a lot of significant features that the Ganglia community is waiting for. I don't expect an x.1.0 version of Ganglia to be completely issue free on all platforms. I fully expect us to release a 3.1.1 fairly soon with more of these types of issues resolved. Just because we officially released Ganglia 3.1 doesn't mean that development or maintenance for lesser common platforms will cease. It also doesn't mean that all new 3.1 functionality will cease either. All it means is that we drew a line in the sand and called it Ganglia 3.1.0. We will draw another even better line in the sand and call that 3.1.1 maybe next month. >> We created the Ganglia 3.1.x stable branch almost 2 months ago and have been > working towards a release since then. > > and in those 2 months we hadn't been able to get enough people interested in > testing this branch to ensure all the designs are sound and that the package > is ready to be going to beta. > Correct, however waiting longer for interested people to show up isn't going to solve the problem. This is open source, we generate interested people by providing new features and functionality. A first major release of any version is not expected to be perfect, that is why there will be minor releases. Ganglia 3.1 may not solve all of the issues on all platform until 3.1.8 or .9 or whatever. But holding back everything that does work just because there are a few issues that are not yet resolved, doesn't do the Ganglia project or the Ganglia community any good. All of these issues will be resolved, it just probably won't happen in the first release. And it definitely won't happen until Ganglia community developers stand up and resolve them. Like I said before, those of us who have been actively working on Ganglia 3.1, have done an excellent job. But there aren't that many of us and we all have day jobs besides working on Ganglia. If someone in the community wants an issue fixed that we haven't be able to resolve yet, then stand up and contribute. Everyone is welcome here. >> I think it is time that we actually create a Ganglia 3.1 release candidate > which can be tested and ultimately determined to be our Ganglia 3.1 official > release. > > did I miss the alphas and the betas? > Why do we need X number of alphas or betas before we release? When the functionality is ready to go, then it is ready to go. Again this is Open Source, a "snapshot", an "alpha", a "Beta" or a "Release Candidate" are all the same thing. They are just source code tarballs that we have released and asked the community to test. We have already done that on a number of occasions and plan on doing it again with an RC1 (or whatever it is called). It doesn't really matter what we call them and the name won't bring anymore interested community testers to the table. We don't have an established "Quality Assurance" department tasked with producing test cases and test plans. I wish we did, but we don't. The Ganglia project has to rely on volunteers just like you and I. With that community of volunteers, we have to do the best we can. But we can't wait forever. The features like the Gmond modular functionality and the improvements in the web frontend have been sitting in the Ga nglia repository for a year or more. How long do we wait to release these features? When do we know that these features have been tested adequately and to everybody's satisfaction? The best way to make that happen is to release the features and let the community tell us what we did wrong or right. Nothing is set in stone here. If somehow we really screwed up 3.1, then let the community tell us, let's fix the problems and move quickly to 3.2. I believe 3.1 is good and ready to go. >> The STATUS file has been cleared out of all backports and there are no > show-stoppers that would prevent us from releasing 3.1. > > the STATUS file shows 1 show stopper, which as you said is not a regression, > but needs to be addressed so that distributions will be able to carry on > with > packages of ganglia 3.1 when it gets out of the door. > No, the issue stated in the STATUS file is not a regression and that didn't stop Ganglia 3.0 from being very successful. Despite being discussed on the mailing list, nobody has stepped up to fix the issue. So until somebody does step up, what do we do? Holding the rest of Ganglia 3.1 back waiting for a previously existing issue to be resolved, doesn't help. > having distributions involved with a beta, will also have the side effect of > making testing more effective by providing a framework and the needed > dependencies. > Not sure what you are suggesting here. Distributions have been involved with the development and testing of Ganglia 3.1. A general release of Ganglia 3.1 is not new news. I have been pushing to get Ganglia 3.1 released for the last six months. Just because we decide to call something a beta, doesn't mean that more distros are going to finally get on board. We aren't releasing Ganglia 3.1 for the various distros benefit, we are releasing Ganglia 3.1 for the Ganglia project's and community benefit. The distros are community members just like the rest of us. If they want to get involved, their participation is more than welcome. >> Unless there is something that should be considered to be a show-stopper, I > propose that we tag and roll Ganglia 3.1 RC1 immediately and start testing. > Once Ganglia 3.1 RC1 tarballs have been made available, then we allow a two > week test period. If no show-stoppers or major issues are found within the > two week period, then at the end of the two week period we officially release > Gangia 3.1. > > ganglia 3.1 should be released when it is ready and unless there is some > time > constrain which I am unaware of. > > rushing a release is not in anyone's interest and will only result in > unhappy > users finding themselves as beta testers of what they expected to be a > stable > release. A well organized beta which will cover all platforms, > distributions > and scenarios we expect to have in production is. > I disagree. Ganglia 3.1.0 is ready. No, not all issues for all platforms are resolved. But that is why there will be a Ganglia 3.1.1 and Ganglia 3.1.2, etc. releases. Again this is Open Source, if somebody finds themselves unhappy because the Ganglia project released a 3.1.0 product that may not be up to their standards, then I fully expect that same unhappy community member to join in the effort and fix the problem. I'm not being paid to fix everybody's individual platform problems and neither is anybody else who actively develops Ganglia. This is an Open Source community effort, not a for-profit commercial product. As such we are all (Ganglia users included) beta testers as well as developers, documentation writers and so on. We all do the best we can and welcome any and all participants to pick up were our best intentions might have failed. Ganglia 3.1.0 is ready to go so let's get it out the door and move on to 3.1.1 Brad ------------------------------------------------------------------------- Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://sourceforge.net/services/buy/index.php _______________________________________________ Ganglia-developers mailing list Ganglia-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ganglia-developers
