On Sat, Jul 09, 2005 at 07:29:10PM -0400, Geir Magnusson Jr. wrote: > > Maybe you could share how things work at OpenEJB? I see that the CVS > appears to be divided into several top-level directories : > > /contrib > /openejb > /openejb-sfnet > /openejb1 > /website > > Are these treated differently? I assume that they have different > release cycles (where appropos...)? Committers have broad access? > > I suspect that the answers are "no", "yes" and "yes" and is generally > how I think our subprojects should work :
We're a bit of a mess right now :) Had we been using svn the whole time, we'd probably have one repo with proper branches. Moving from exolab->sf.net->codehaus didn't help. Moving the website out was a big mistake as the docs just got really out of sync with the different versions, Jacek will probably be moving them back. I'm kicking myself in the butt for thinking starting a new cvs repo for each major version was a good idea. And don't even get me started on releases. Anyway, I digress.... I think OpenEJB has managed the growth of a flood of new code and committers about the worst that anyone could :) I don't know that I'd want to model anything we've done in that regard. > 1) Distinct partitioning of the subprojects > 2) Broad committer access to enable a strong community, with the > community value that when entering a new area, be respectful of > what's there, ask questions when you don't understand, etc... > 3) Separate release cycles so that the parts are easy to depend on > both within the project and without > 4) Single PMC overview and responsibility Items 2, 3 and 4 look good. I'm not sure what disctinct partitioning entails. > For example, I could see us doing a [....] I'd like to see us maybe do these someday if we could: - J2CA subproject (Jencks :) - Security subproject - ORB subproject -David
