On Sun, 2006-04-09 at 19:19 -0400, Noel J. Bergman wrote: > Andrew C. Oliver wrote: > > Then there is no NEED for a sandbox. > > As you know, the sandbox predates the Incubator, and AIUI, the Sandbox > exists so as to allow experiments without polluting the respository in such > manner that would confuse the public and ourselves about what is real and > what is play. There may be other ways in to achieve that goal.
Agreed. I think much of the Sandbox concept owes its existence to the limitations of CVS, and that with Subversion and the recent jakarta-wide commit access a lot of the need for a sandbox is gone. A project which has ties to an existing one (eg a refactoring of common code out of a project into a "common component") can be done in a sandbox subdir of that project (sibling to trunk/tags/branches). Discussion can be held on that project's lists. Oversight is provided by the committers on that related project. When it's ready to be promoted, a simple "svn mv" and the creation of a separate email list will do the job. For projects which are brand new but likely to become part of jakarta commons, the existing commons sandbox (using the existing commons-dev list) seems appropriate to me. Oversight is provided by the commons community. Of course if the project is a kind of "language extension" then it might want to hang out on the proposed commons-lang-components list instead of the original commons list. Projects that originate outside apache and are being brought in go through the incubator of course. Oversight is provided by those kind apache committers who subscribe to the incubator lists. The only problem I see is largish projects that are originated by existing Apache committers and have no close affiliation to existing projects. There aren't likely to be very many of these. I would suggest that if such a project can't find an existing project willing to effectively "sponsor" it by allowing their own list and subversion dir to be used to host the project for a while, then it belongs in incubation. The other issue to consider is where websites for sandbox-status projects can live. I think it would be nice to group these together, eg under jakarta.apache.org/sandbox. This provides a way for such projects to publish sites while making it clear to users that they aren't yet "approved". To summarise: I suggest setting up a parent website for jakarta-wide sandbox stuff, and dropping the existing sandbox docs that encourage non-commons projects to come and play in the commons sandbox. Otherwise things can be pretty much left as they are... Cheers, Simon --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
