Hey hey, Ooh, interesting topic. I'm going to top-post because I want to focus on the big picture!
* Apache HTTPD provides 8+ years of support for old releases. * Tomcat provides 6+ years of support for N-2 release. * Ant provides 12+ years of backward compatiblity, so far. (details below) I think this is great and when I was proud of apache it was usually because of stuff like that. Every now and then I get a support enquiry about code that has been in the attic for many many years, and I always take the time to answer it, even if I've almost forgotten about collections pre java 1.2. This loooong term support happens because the people that work on those projects want it to happen and do the work to make it happen. Since in this case, you want it to happen, and signed up to do the work, cloudstack its support window (for 4.3) grows. The more people do that, the bigger the support window will get. The 4.3 branch should live as long as people want to work on it and there's enough people to vote to release it. No-one should stop you, and I'd be a little upset if someone tried. This can happen naturally: it doesn't actually *need* a model or a discussion, just people to do the work and enough people to vote to release that work. You see a need here, you're stepping in to fill that need, so, "thanks for volunteering" (no sarcasm). I personally believe such explicit support models and commitments can hurt for 'upstreams' (*). If you look at the httpd download page, it doesn't say "we'll support this for 8 years to come", it just says 'download here'. Users are expected and trusted to evaluate whether the community support is enough, and if it isn't, or they can't figure that out, they should go seek a downstream that provides the support (and typically, warranty and guarantee and indemnification and SLA and ...) that you don't get from an open source project. Ubuntu is a differently shaped project from cloudstack. Ubuntu is a (more unstable...) downstream of debian, where the httpd package is a downstream of httpd.apache.org. The key value of ubuntu LTS is in the tested _aggregation_ of many mutually compatible versions. IMHO. But hey, agreement is absolutely not required! I applaud you for doing what you think is right for your customers and for talking openly about it here. Customers these days tend to be pretty good at spotting who is listening to what they need, so as long as you understood that correctly, I'm sure it's a sound commercial decision for ShapeBlue too :-D cheers, Leo (*) I think in the loooong term that quality improvement is best focused on master/tip. Well, at least up to about 80% unit test coverage or so :). My advice would be to ditch all 4.3 work, ditch any further 4.4 work, and invest all that effort into /testing/ for 4.5. Once you have high code velocity, trustable continuous integration and continuous delivery, etc, compatibility&stability are just more things to test&measure, and they only go up. ------ HTTPD * 2002-02-06 first release of apache httpd 2.0 * 2002-02-03 last release of apache httpd 1.3 That's a history of 8 years of support for N-1 major releases. * 2005-11-30 first release of apache httpd 2.2 * 2012-02-19 first release of apache httpd 2.4 * 2013-07-02 last release of apache httpd 2.0 That's a history of 8 years of support for N-1 minor releases. * 2.2 and 2.4 currently still being supported So so far that's 9 years of support for the current N-1 minor release. Of course httpd 2.4 is ~99% backward compatible with httpd 2.0, so that's 12+ years of backwards compatibility. Tomcat * 2004-08-29 first releaes of tomcat 5 * 2006-10-21 first release of tomcat 6 (still supported) * 2011-03-05 first release of tomcat 7 (still supported) * 2012-10-09 last release of tomcat 5 * 2014-02-02 first release of tomcat 8 So that's a history of 6 years of support for N-2 major releases. Ant * 2003-08-12 first release of ant 1.5 (1.5.2) * 2014-04-30 current release of ant (1.9.4) Ant's been ~99% backward compatible from about ant 1.4, but I can't find a timestamp for ant 1.4. So that's a history of 12 years of backward compatibility.
