L.S.,
I don't think that the installation (disk space) footprint would change a lot, to be honest. For the default assembly, that already doesn't include any of the JBI/NMR stuff, so it is just Karaf + Camel/CXF/ActiveMQ JARs. The JBI assembly would disappear and the full assembly would probably become only slightly smaller (because we drop the NMR/JBI + JBI components, but a lot of their dependencies are part of Camel & co as well). The whole point of the full assembly is to have something that contains all the runtime dependencies for offline use, so that's bound to be a rather large distribution. We added the minimal assembly for people that don't like our default dependency set (Camel, CXF and ActiveMQ) and just need one or two of those, so they can start with an empty container and only add the things they want. What particular combination of components/dependencies did you have in mind when you were thinking about a smaller footprint assembly? Regards, Gert Vanthienen On Mon, Feb 4, 2013 at 12:17 PM, Corrado Campisano < [email protected]> wrote: > hi > > will the choice to "remove everything that's related to JBI and the NMR" > will make the installation footprint better? > > Karaf is 10MB (as much as smix4 minimal assembly), other smix4 assemblies > are default=50MB, JBI=80MB, full=200MB (as much as Fuse ESB is), what about > smix5 assemblies? > > > Regards, > corrado > > > 2013/2/4 Willem Jiang <[email protected]> > > > It sounds good. Keep it simple and deliverable is a best practice we > > always do :) > > > > 发自我的 iPhone > > > > 在 2013-2-4，上午4:54，Gert Vanthienen <[email protected]> 写道： > > > > > L.S., > > > > > > > > > About a year and a half ago, we had some discussions on the mailing > list > > > about a plan for Apache ServiceMix 5.0 and had some initial commits to > > > build the additional services and functionality. Since then however, > > none > > > of us have actually had the time to work on that code or move things > > > forward. > > > > > > In the meanwhile, we are also struggling constantly to get our releases > > > done in timely fashion. The latest 4.5.0 release took almost 9 months > > > since the first mention of it on the dev@ list. Doing a ServiceMix > > release > > > now is quite a task: it usually involves doing releases in 5 or 6 > > > subprojects. > > > > > > I would like to propose a new plan for Apache ServiceMix 5.0. Instead > of > > > doing a lot of new development, how about we start with the current 4.x > > > features codebase and remove everything that's related to JBI and the > > NMR. > > > That will give us a nice and simple integration container build (based > on > > > Karaf, Camel, CXF, ActiveMQ, ...) and everything is living in a single > > > project that's quick and easy to release. > > > > > > If we start doing this now, we could get a build out with Karaf 2.3.0, > > > ActiveMQ 5.8.0, Camel 2.11.0 (which will bring in Scala 2.10 and opens > up > > > the possibility to include the Akka OSGi examples I built a few months > > ago) > > > pretty soon after those versions are available. With only one project > > to > > > maintain the versions of all those dependencies, we should be able to > > > follow up more regularly as our sibling projects do (new) fix releases > as > > > well. > > > > > > We don't have to throw away the existing ServiceMix 5.0 code by the > way, > > we > > > can always move that into a separate branch and then cherry-pick the > > useful > > > bits afterwards, but I think our first goal now should be to get > > ourselves > > > in a position that we can actually build and release stuff more easily > > > again. > > > > > > > > > Wdyt? > > > > > > Gert Vanthienen > > >
