On Jan 27, 2008 11:25 AM, Peter Tribble <peter.tribble at gmail.com> wrote: > On Jan 20, 2008 8:53 PM, Shawn Walker <swalker at opensolaris.org> wrote: > > > > Based on comments received so far, I have completed a new version of > > this proposal. That nature of the proposal has not changed, although > > it should be much clearer what is being proposed now. > > I'm against this, as it tries to take a broken and divisive situation (the > question of the OpenSolaris distribution) and turn the current state into > official policy; and does so using the current constitutional mechanism > which is also in need of redefinition.
It really does no such thing. I swear. All the proposal is about is creating a Distribution CG where distribution projects can be properly sponsored. That's it. The proposal does not attempt to define what an "OpenSolaris distribution" should be. It does not define any specific policy either. > I'm not saying that the general thrust is wrong. But it really needs to be > preceded by an attempt to answer the basic questions - what is OpenSolaris? > What are we trying to achieve? And how are we going to do that? I think that's well outside the scope of this proposal. > > Summary > > ========== > > The OpenSolaris community needs a new Community Group to facilitate > > collaboration, design, and development of OpenSolaris-based > > distributions. > > Does it? I fail to see how any distribution needs a CG to oversee its work. Because a distribution is a project (only projects get repositories, etc.), and projects according to the constitution, need a CG to oversee them. > > Because of new tools that are already available to the > > community, but are rapidly reaching maturity, it is probable that > > there will be a large number of community-based distributions being > > created soon. > > In which case that's a problem, and something to be avoided rather > than encouraged. One size fits all has never seemed to work with any open source operating system project; I don't see how OpenSolaris can succeed any differently. This isn't so much about encouraging people to start their own as it is providing a place for them to collaborate on existing ones. The community has already indirectly spawned Nexenta, Belenix, and SchilliX. Isn't it better to have a central place *here* where the community can actively participate in distribution projects like those then having that innovation occur outside our community (i.e. opensolaris.org)? > (And if you're thinking of the Distro Constructor, then I don't regard that > as terribly well named - it's more constructing custom install images > [akin to jumpstart] around a single distro than building independent > distros.) I will have to disagree. > > In addition, the current set of Community Groups > > available to sponsor projects are not properly scoped or otherwise > > suitable for the nature and goals of distribution projects. As such, > > the creation of a new Community Group is needed. > > Is this because distributions don't fit neatly into the constitutional > model? In which case, creating another CG doesn't help. It's not > at all obvious to me that all distributions would want to be under > the direction of a CG (independence is one virtue I would want), > or that more than one distribution can exist under those terms. My belief is that more than one distribution can exist. This proposal is just that; a proposal. I have no idea if it is appropriate or not, that's why it's a motion to the OGB :) Regardless; I think it is quite apparent that there is a need for a Distribution CG; whether the current definition and/or structure of a CG fits properly to the proposal is another story. However, at this time, it is the only governance structure available to me for this proposal. > > Scope > > ========== > > The proposed Distribution Community Group shall be the group to which > > the OGB delegates responsibility for encouraging the sustained growth > > and success of OpenSolaris-based distribution projects. > > That sounds like advocacy to me. I suspect, though, that you're > implying something more technical? Yes and no. A distribution project inherently has advocacy and technical aspects. > > Distributions > > that have previously been sponsored by another Community Group may > > remain with that group. However, it is hoped that all such projects > > will approach the contributors of this new community to seek > > reassignment (sponsorship), or that if necessary, they be reassigned. > > What benefits would a distribution have by virtue of being sponsored > by the proposed CG? Based on comments of others, the current CGs available are not appropriate for sponsorship of distribution projects. The new CG provides an appropriate home. > > The Distribution Community Group should be the central place for > > discussions and decisions regarding OpenSolaris-based distributions > > and their impact upon the OpenSolaris community. To support this, it > > is proposed that the OGB will allow the group to act as the initial > > arbiter in dispute resolution amongst distributions and related > > projects. > > This makes no sense. If there is more than one distribution, then each is > independent, and responsible for its own decision making. If there's only > one then there will not be disputes, that distribution would be answerable > to the community as a whole, and the layer of a CG is unnecessary - you > may as well make the distribution its own CG. Which may just have to happen; but I hope not. To me, the last thing we want is fragmentation among the distributions created by our community. Creating a more closely-knit group that allows projects freedom while trying to ensure that they work with other CGs as appropriate seems sensible. > > Some examples of dispute resolution might include: acting as an > > arbiter between distribution project leaders when disagreement between > > them arises, > > Why would they disagree? (Apart from the fact that they will, by virtue of > being separate distributions, have different goals.) And if two distributions > disagree as to how to proceed, then why does that matter? Because a healthy community, will have disagreement sometimes, and that should be expressed in a constructive manner. I'd rather not try to invent hypothetical situations to justify this. It just seems logical to me. If you look at the various BSD distributions, GNU/Linux dsitributions, etc. out there today; you will find disagreement has occurred between them. I just feel that having a Distribution CG oversee the distribution projects makes it easier to ensure that a project is working with other, appropriate CGs and that conflict is prevented rather than resolved when possible. > > ensuring that any relevant policies and guidelines > > established by the constitution or authorised governing body are > > followed, > > Why wouldn't they be anyway? You don't need a CG for this. You do actually. Projects can't exist without a CG. > > and encouraging reconciliation when there are disputes. > > > > The Distribution Community Group will actively work to ensure that > > there is a healthy relationship between its sponsored projects and the > > rest of the community. It will accomplish this by doing things such > > as: ensuring that feedback from the group's sponsored projects is > > provided to the related Community Groups and projects, encouraging the > > further development and creation of tools related to the development > > of distribution projects, and ensuring that projects coordinate their > > releases with related Community Groups (such as the Advocacy Community > > Group). > > > > The Distribution Community Group will also work to establish unified > > processes and resources to assist distribution projects in > > contributing to the sustained growth and success of our community. The > > usage of a common set of resources will help keep our community's > > efforts from becoming fragmented. Promoting the use of these unified > > resources and processes (such as a reference technology platform) will > > allow individuals to be free to innovate in many areas of OpenSolaris > > technology instead of reinventing the basic tools and processes that > > every distribution project needs to operate. > > That actually sounds like a bunch of projects, rather than something > that a CG would do (CGs are just managerial artifices). Yes, and many of our CGs today have a "bunch of projects." I'm not sure I follow you here. As Ben pointed out; projects aren't defined by the constitution currently. This CG is way to group those projects. > > Initial Projects > > ========== > > The following are a series of projects that the Community Group should > > begin once it is formed. > > > > With the acceptance of Project Indiana, and the acceptance of the > > initial set of core contributors, Project Indiana should be reassigned > > to the new Community Group or seek re-sponsorship. > > > > As part of encouraging the sustained growth and success of > > community-based distributions, it is highly desirable that a new > > project oriented towards ON Community Group developers be created. > > This new project would maintain a branch of the main ON tree that > > integrates patches from community developers on a rapid basis as they > > are approved for inclusion. It will build the resulting source tree on > > frequent basis (to be determined, with the initial goal being weekly). > > This will allow community developers to quickly see and test the > > results of their contributions while encouraging innovation and > > providing an easy method for developers to obtain feedback from other > > community members. > > Where does this fit into the creation of a Distribution CG? I fail to see > the relation, as either could be done completely independently of > the other. In fact, if creation of an unstable tree is proposed, then it > would be better done separately . > > I don't see this as a good idea in any case. It requires an extra level > on management, and having two trees just creates confusion. I don't > see what problems it's trying to solve that current efforts to open up the > existing ON tree do not address. As should have been clear, this section of the proposal was merely background or informative. It is not actually what is being proposed to the OGB. This section has been removed in the latest version to reflect that. > > The aforementioned project, will also serve as a testbed for > > developing unified processes and tool usage (such as the distribution > > constructor) while ensuring individuals are free to innovate. The > > resulting distribution, by using these tools and processes, will be a > > valuable asset. > > > > Rationale > > ========== > > There are many reasons a Distribution Community Group is needed. One > > of them is that past events have shown that none of the existing > > community groups are wholly suitable for the community-wide impact > > that distributions can have within the OpenSolaris community. In > > addition, none of the existing Community Groups are properly scoped > > for the kind of decisions that need to made to by a knowledgeable, > > focused group of individuals whose primary interest relates to the > > production of distributions. > > > > Additionally, the Distribution Community Group will help establish a > > clear chain of responsibility by ensuring that distribution projects > > are directly responsible to a particular Community Group. This allows > > a first line of arbitration, and active guidance that distribution > > projects need. > > > > Finally, the tools necessary to build a distribution are quickly > > reaching maturity, and the Community Group will encourage individuals > > creating or maintaining distributions to participate directly in the > > OpenSolaris community to avoid some of the past problems that have > > arisen. > > > > [Proposal Version 3 - Updated Jan 19th, 2008 12:35am CDT] Thanks for your feedback Peter. However, what we have today does not work. Various distribution projects were started somewhere other than opensolaris.org for whatever reason. At least a few individuals felt that distribution projects, such as Indiana, are not suposed to be sponsored by CGs such as Desktop. What CG then would be appropriate for distribution projects to you? How can we encourage developers to start their distribution projects here? It is not a question of if people will create their own OpenSolaris distributions; they have and will continue to do so. The question, how would you seek to guide them in such a way that their innovations will be re-incorporated for the benefit of the entire community while discouraging fragmentation? Cheers, -- Shawn Walker, Software and Systems Analyst http://binarycrusader.blogspot.com/ "To err is human -- and to blame it on a computer is even more so." - Robert Orben
