Bryan Cantrill wrote: >> My test is this: if someone were to go to opensolaris-discuss and >> propose DTrace or mdb now - would it be classified as a project, or as a >> community? If someone proposed another debugger, let's say "ndb", or >> someone proposed to port (god forbid) SystemTap to Solaris (ignore GPL >> issues, this is hypothetical) - would we grant them a community? No, we >> would say it's a project. >> >> So this seems to suggest that we need a path for "promoting" projects to >> be communities once they have reached some threshold. > > I agree. In fact, part of the reason that I _don't_ want DTrace to > be in the Observability community is that in the presence of say, a > SystemTap port to Solaris (shudder), it makes absolutely no sense for > SystemTap and DTrace to be in the same Community -- because they're > very much _not_ in the same community. (A more current example would > be ZFS and QFS, the communities of which are largely orthogonal.)
... but then you could arrange community food fights, and you have to admit there's some value in that. >> So maybe we give up the idea of uniformity in classification of >> communities/projects - and say *everything* starts as a project until it >> reaches said threshold at which point it becomes a community with >> governance? It lends itself to more organic growth, which seems to be >> the feedback I've been receiving here. > > Organic is good -- you don't want to be too rigid about the definition > of Community and Project. That said, there _is_ a distinction in the > Constitution and it's unfortunate that we have overloaded the term > "community" with a governance definition. It means that you either have > to (1) give every self-organizing community the right to determine the > fate of OpenSolaris or (2) tell some self-organizing community that, > while they might be a community, they are not, in fact, a Community. > Both paths are problematic, though if one must choose only between these > stark choices are would rather empower too many (overly inclusive) than > empower too few (overly exclusive). So having thought about it, governance should be given to "the people". So as projects grow and accumulate "momentum" to the point at which they are a community (whether that's DTrace, Whizzix - the newest OpenSolaris distribution, or the "UFS Support for Vista" project) - maybe they should be given representation. We would still have the conflict between "SystemTap for OpenSolaris" (if it got big enough) and "DTrace", but at least it would be based on representation rather than classification. > But I think the notion of starting everything out as a project and > promoting accordingly is a good one. And I would also recommend a third > moniker for user groups, which I think is a large part of the governance Agreed > problem (like, say, User Group), but I imagine that that will require > constitutional modifications... sod that... it would require webapp modifications (which are possibly more painful than constitutional modifications, sadly) Thanks for the feedback Bryan cheers, steve (with thanks to Stephen Hahn for some enlightening conversation) -- stephen lau // stevel at sun.com | 650.786.0845 | http://whacked.net opensolaris // solaris kernel development
