From Eric.Boutilier at Sun.COM Wed Oct 4 06:21:54 2006 From: Eric.Boutilier at Sun.COM (Eric Boutilier) Date: Wed, 04 Oct 2006 08:21:54 -0500 (CDT) Subject: [cab-discuss] Re: [osol-discuss] Community Proposal: HPC Developer In-Reply-To: <4277665.1159547054264.JavaMail.Twebapp@oss-app1> References: <4277665.1159547054264.JavaMail.Twebapp@oss-app1> Message-ID: Brad, Note that because this is a community, not project, proposal, you'll need at least one CAB member to give it a +1. I've copied them on this message. Eric On Fri, 29 Sep 2006, Brad Lewis wrote: > I propose adding an OpenSolaris HPC Developer Community. The goal of this community > is to promote the use of OpenSolaris in High Performance Computing by providing tools, > libraries, and utilities for developers on OpenSolaris. The scope will span shared and > distributed memory development. > > I'm in the Sun Performance Library group. We have several possible projects that may be > of interest to the HPC Community. Here are a few examples. > > CBLAS for Solaris - The netlib cblas package does not work correctly with SunStudio > compilers. The SunStudio mixed language programming model requires string lengths > to be added to the end of the argument list for all string arguments when C calls Fortran. > We would like to provide a CBLAS with these arguments. > > CORD(COurse Region Data flow) API - This API provides capability similar to OpenMP > work queues only more powerful. Data dependencies are specified between tasks. This allows developers to more fully exploit parallelism in algorithms that are more complicated > than a set of independent tasks. > > AlgoVis (Algorithm Visualizer) - AlgoVis records memory references during an application > execution. It then maps these references to Matrix and Vector objects and displays a movie > of the execution to help developers visual algorithms. > > Iterative Solvers -provide iterative methods for common matrices, symmetric matrices, large and > ill-conditioned systems, etc. > Common Numerical Library - Common Numerical computations outside of linear algebra including > differential equations and systems, stiff systems, approximation, polynomial interpolation, spline > approximation, integration, > > Cluster Library - Provide parallel distributed computation for clusters. > > > There is a large interest in HPC Development inside and outside of Sun. Internal to Sun, the HPC > interest discussion list boasts 289 members. Existing open source interest is demonstrated by > projects such as LinuxHPC.org which claims to have 750+ visitors daily. > > There may be some overlap with the performance community in regard to some aspects of > performance tuning. > > Thanks, > > Brad Lewis > > > This message posted from opensolaris.org > From rich.teer at rite-group.com Wed Oct 4 08:15:05 2006 From: rich.teer at rite-group.com (Rich Teer) Date: Wed, 04 Oct 2006 08:15:05 -0700 (PDT) Subject: [cab-discuss] Re: [osol-discuss] Community Proposal: HPC Developer In-Reply-To: References: <4277665.1159547054264.JavaMail.Twebapp@oss-app1> Message-ID: On Wed, 4 Oct 2006, Eric Boutilier wrote: > Brad, > > Note that because this is a community, not project, proposal, you'll need > at least one CAB member to give it a +1. I've copied them on this message. This isn't a +1/-1 vote from me, but a request for more info. Given the proliferation of Communities at the moment (and the fact that we're trying to reduce the number Communities to a more manageable number), is there a need for a separate HPC Developer Community? Could it's needs be met as, say, a sub-Community (as it were) of the Developer Community? -- Rich Teer, SCNA, SCSA, OpenSolaris CAB member President, Rite Online Inc. Voice: +1 (250) 979-1638 URL: http://www.rite-group.com/rich From al at logical-approach.com Wed Oct 4 08:16:34 2006 From: al at logical-approach.com (Al Hopper) Date: Wed, 4 Oct 2006 10:16:34 -0500 (CDT) Subject: [cab-discuss] Re: [osol-discuss] Community Proposal: HPC Developer In-Reply-To: References: <4277665.1159547054264.JavaMail.Twebapp@oss-app1> Message-ID: On Wed, 4 Oct 2006, Rich Teer wrote: > On Wed, 4 Oct 2006, Eric Boutilier wrote: > > > Brad, > > > > Note that because this is a community, not project, proposal, you'll need > > at least one CAB member to give it a +1. I've copied them on this message. > > This isn't a +1/-1 vote from me, but a request for more info. Given > the proliferation of Communities at the moment (and the fact that > we're trying to reduce the number Communities to a more manageable > number), is there a need for a separate HPC Developer Community? > Could it's needs be met as, say, a sub-Community (as it were) of > the Developer Community? Or a separate Project? Al Hopper Logical Approach Inc, Plano, TX. al at logical-approach.com Voice: 972.379.2133 Fax: 972.379.2134 Timezone: US CDT OpenSolaris.Org Community Advisory Board (CAB) Member - Apr 2005 OpenSolaris Governing Board (OGB) Member - Feb 2006 From Brad.Lewis at Sun.COM Wed Oct 4 10:00:52 2006 From: Brad.Lewis at Sun.COM (Brad Lewis) Date: Wed, 04 Oct 2006 11:00:52 -0600 Subject: [cab-discuss] Re: [osol-discuss] Community Proposal: HPC Developer In-Reply-To: References: <4277665.1159547054264.JavaMail.Twebapp@oss-app1> Message-ID: <4523E8C4.3040204@sun.com> Let me explain more why I proposed a new community. I started out thinking about starting a project. My interest is in parallel programming and I have developed an API for data flow parallel programming for linear algebra in Fortran. Given the fairly narrow audience for that I was trying to determine if that was an appropriate OpenSolaris project. As I talked to OpenSolaris folks I knew most seemed to think that yes it would fit. Some suggested that it was in fact needed because there were not anything really happening on the traditional HPC scientific computing front. As I planned on moving forward to start a new project I discovered that there was quite a bit of interest within my own working group and closely associated groups in starting there own projects centered around providing support for HPC developers. I asked around the OpenSolaris Performance Community to see if these projects would fit with what they are doing and I came to conclusion that the focus there is different. Scientific and technical computing is a very different domain than the kernel. That's when I starting thinking about proposing a new community. I think an HPC community will attract a different set of developers. It seems a new community is a better answer than trying to hijack an existing on. If the community can develop some good support I think OpenSolaris clusters could really take off. A lot of HPC research is being done on "cheap" hardware and linux clusters. I also think are some opportunities for HPC on upcoming CMT systems. Al Hopper wrote: >On Wed, 4 Oct 2006, Rich Teer wrote: > > > >>On Wed, 4 Oct 2006, Eric Boutilier wrote: >> >> >> >>>Brad, >>> >>>Note that because this is a community, not project, proposal, you'll need >>>at least one CAB member to give it a +1. I've copied them on this message. >>> >>> >>This isn't a +1/-1 vote from me, but a request for more info. Given >>the proliferation of Communities at the moment (and the fact that >>we're trying to reduce the number Communities to a more manageable >>number), is there a need for a separate HPC Developer Community? >>Could it's needs be met as, say, a sub-Community (as it were) of >>the Developer Community? >> >> > >Or a separate Project? > >Al Hopper Logical Approach Inc, Plano, TX. al at logical-approach.com > Voice: 972.379.2133 Fax: 972.379.2134 Timezone: US CDT >OpenSolaris.Org Community Advisory Board (CAB) Member - Apr 2005 > OpenSolaris Governing Board (OGB) Member - Feb 2006 > > From sch at eng.sun.com Wed Oct 11 12:21:19 2006 From: sch at eng.sun.com (Stephen Hahn) Date: Wed, 11 Oct 2006 12:21:19 -0700 Subject: [cab-discuss] Review comments on Article 3 In-Reply-To: References: <20060802055515.GA134836@eng.sun.com> <55009E9F-D5E9-42DE-92A8-AA1F8F31D328@gbiv.com> Message-ID: <20061011192119.GB8499@eng.sun.com> * Simon Phipps [2006-09-27 12:59]: > > On Sep 27, 2006, at 20:40, Roy T. Fielding wrote: > > >On Aug 1, 2006, at 10:55 PM, Stephen Hahn wrote: > > > >> Small corrections: > >> > >> 3.2 "by applicable law or regulation" -> "by applicable law, > >>rule or > >> regulation" > > > >I made this change, but it still looks odd to me. Is that supposed > >to be "rulings" instead of "rules"? "rules" is just too broad a term > >to be subject to in general. > > Yes, I'd like to understand what extra context this allows for as well. I was unfortunately only relaying the comment; I had assumed it to be boilerplate. - Stephen -- Stephen Hahn, PhD Solaris Kernel Development, Sun Microsystems stephen.hahn at sun.com http://blogs.sun.com/sch/ From Glynn.Foster at Sun.COM Mon Oct 16 03:07:01 2006 From: Glynn.Foster at Sun.COM (Glynn Foster) Date: Mon, 16 Oct 2006 23:07:01 +1300 Subject: [cab-discuss] Re: Governance Draft [was 'Growing the OpenSolaris Community'] In-Reply-To: <82693572-7FC4-427F-B94B-61DBCBEF2241@sun.com> References: <452DCDBC.9030903@Sun.COM> <4532EF01.9090707@sun.com> <82693572-7FC4-427F-B94B-61DBCBEF2241@sun.com> Message-ID: <453359C5.70503@sun.com> Simon Phipps wrote: > On Oct 15, 2006, at 19:31, Glynn Foster wrote: > >> o CAB formed 1 year+ ago, yet still no governance model, or idea >> of more formal membership for the project > > > Actually there is a comprehensive and well-thought-through governance > model in existence, which the CAB has been working on with little > community feedback - see http://www.genunix.org/wiki/index.php/ > OpenSolaris_Governance_Draft_02 for which feedback is still sought. > The CAB is in the final stages of agreeing this version. Almost certainly a little too late, but I thought I'd spend a couple of minutes reading over the latest draft on the back of this mail. In general it reads pretty well. I'm not personally a big fan of over complicated governance models, but I can appreciate the need to having a little more structured and formal in this case. Glynn ARTICLE I. Name I don't like 'virtual' - I understand the need for it, but I don't like it ;) ARTICLE IIIa. Structure, Participation, and Roles I do have concerns that we're setting the bar a little high that only Core Contributors can become [voting] members. In GNOME, anyone who shows a commitment and non-trivial improvement to the project gets to become a Foundation member with voting rights - anything we can do to encourage people to get involved is a step in the right direction. Voting typically means being able to vote in the annual Foundation Board elections, rather than technical decisions. I would assume in this case, Core Contributors would vote on a per-project, per-community basis - in that, only CC's that come from that particular project/community get to vote. However, all CC's can vote on opensolaris.org wide decisions? ARTICLE IV. Meetings of Members I wonder if classifying that meetings can only be held via a shared electronic communication medium is a good idea. While I firmly agree to it if there is any issues that need to be voted on, I think it really limits future possibilities of meetings at conferences or otherwise. I don't like the idea of Voting Proxies. It only complicates things in my opinion. I prefer the requirement [as detailed in a later section] for a voting timeline, allowing members to get their votes in over the course of a couple of days. If they are not around, or not provided an alternative contact address for themselves, they lose out. ARTICLE V. Governing Board I'm somewhat surprised that OGB calls are open to Members. Surprised in a good way, although I wonder if the content may include confidential information especially if the calls are approximately spaced every 3 months. ARTICLE VI. Community Groups 6.4.5 reads a little strangely, and a little hard to understand. In 6.6, it states that the OGB decides on the list of Core Contributors for an initiated group, yet 6.4 says that the nominator provides this. In general, this feels like a lot of paperwork for some groups to handle, yet for some groups, this process may not map very well [though that may be a reflection of 'failed' groups that we currently have]. I would suggest changing 'In addition, each Community Group shall be assigned an archived private mailing list for limited use by the...' to 'may optionally be assigned'. I don't think we necessary want to encourage the use of private mailing lists. ARTICLE VII. Community Group Voting Procedures I have no issues with the voting procedure, although I wonder if the use in casual mailing list conversations will confuse people - especially those who may not have Member voting rights who wish to get involved in the conversation. Some clarification may be needed. Glynn From joshhurst at gmail.com Wed Oct 18 07:53:13 2006 From: joshhurst at gmail.com (Josh Hurst) Date: Wed, 18 Oct 2006 16:53:13 +0200 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: Message-ID: On 10/18/06, Josh Hurst wrote: > On 10/18/06, James Carlson wrote: > > Josh Hurst writes: > > > On 10/17/06, maoyi wrote: > > > > I want to post my 4over6 code.How? > > > > > > > > 4over6 is a project that makes ipv4 packets go through ipv6 network. > > > I suggest that you wait until the ksh93 integration project is done > > > with it's work. The project suffers from excessive bureaucracy > > > overhead and in my opinion a discussion should be started to > > > streamline the project integration process. The current process has > > > become unbearable and this needs to be corrected before new projects > > > are launched > > > > Your fear-mongering is completely misplaced just plain unacceptable. > Please understand that I am trying to start a generalised discussion > whether all this bureaucracy is REQUIRED. No other Open Source project > requires this kind of excessive bureaucracy overhead. > > I do not feel very well that I have to kickstart this discussion but I > think the excessive delays in the ksh93 integration project are no > longer bearable and require at least a look at the general process. > > I'd like to start a constructive discussion about the problems because > a reform is overdue I've added the CAB discussion list to the CC: for comments. IMO the problem should be discussed before any new projects are started. We need a new, streamlined process which makes sure that projects are no longer trashed with excessive bureaucracy -- Josh From james.d.carlson at sun.com Wed Oct 18 08:06:37 2006 From: james.d.carlson at sun.com (James Carlson) Date: Wed, 18 Oct 2006 11:06:37 -0400 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: Message-ID: <17718.17149.27108.150264@gargle.gargle.HOWL> Josh Hurst writes: > I've added the CAB discussion list to the CC: for comments. IMO the > problem should be discussed before any new projects are started. We > need a new, streamlined process which makes sure that projects are no > longer trashed with excessive bureaucracy I must insist once again that you do NOT include the original poster (maoyi ) on your messages. This poster's proposed 4over6 networking project has nothing whatsoever to do with the merits of your complaints about the development process. By directing your comments to the wrong forums and people, all that you are doing is reducing the signal-to-noise ratio and spreading unwelcome (and I believe completely unwarranted) fear of the Open Solaris community. Please do not do that. -- James Carlson, KISS Network Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 From alan.coopersmith at sun.com Wed Oct 18 08:05:26 2006 From: alan.coopersmith at sun.com (Alan Coopersmith) Date: Wed, 18 Oct 2006 08:05:26 -0700 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: Message-ID: <453642B6.1060902@sun.com> Josh Hurst wrote: > I've added the CAB discussion list to the CC: for comments. IMO the > problem should be discussed before any new projects are started. We > need a new, streamlined process which makes sure that projects are no > longer trashed with excessive bureaucracy The "excessive bureaucracy" of the ARC is how Sun has made it possible for Solaris to define what interfaces can be depended on by applications and thus to allow binary compatibility for over a decade. It's also what helps keep projects aligned with each other to deliver a more coherent whole. These are some of the major reasons people prefer Solaris over other OS'es - if you got rid of them, why would people still want Solaris instead of the OS'es already run that way? (This month's ACM Queue magazine has a nice bit of perspective as to how the ARC processes came about - see the pointer on: http://blogs.sun.com/pschow/entry/solaris_evolution_and_interface_stability ) -- -Alan Coopersmith- alan.coopersmith at sun.com Sun Microsystems, Inc. - X Window System Engineering From Casper.Dik at Sun.COM Wed Oct 18 08:08:46 2006 From: Casper.Dik at Sun.COM (Casper.Dik at Sun.COM) Date: Wed, 18 Oct 2006 17:08:46 +0200 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: Message-ID: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> >On 10/18/06, Josh Hurst wrote: >> On 10/18/06, James Carlson wrote: >> > Josh Hurst writes: >> > > On 10/17/06, maoyi wrote: >> > > > I want to post my 4over6 code.How? >> > > > >> > > > 4over6 is a project that makes ipv4 packets go through ipv6 network. >> > > I suggest that you wait until the ksh93 integration project is done >> > > with it's work. The project suffers from excessive bureaucracy >> > > overhead and in my opinion a discussion should be started to >> > > streamline the project integration process. The current process has >> > > become unbearable and this needs to be corrected before new projects >> > > are launched >> > >> > Your fear-mongering is completely misplaced just plain unacceptable. >> Please understand that I am trying to start a generalised discussion >> whether all this bureaucracy is REQUIRED. No other Open Source project >> requires this kind of excessive bureaucracy overhead. >> >> I do not feel very well that I have to kickstart this discussion but I >> think the excessive delays in the ksh93 integration project are no >> longer bearable and require at least a look at the general process. >> >> I'd like to start a constructive discussion about the problems because >> a reform is overdue >I've added the CAB discussion list to the CC: for comments. IMO the >problem should be discussed before any new projects are started. We >need a new, streamlined process which makes sure that projects are no >longer trashed with excessive bureaucracy This is assuming that there is a problem; three CAB members did take part in the development process discussions (Al, Rich and me) and I don't think we so fault in it. Of course, we're all Solaris/Sun fanboys and believe that the way in which we architect and develop is "The One True Way(TM)". The Solaris architecture meetings take some getting used to, I grant you that; and my reaction when first confronted with them was pretty much like yours. "It's my project and I cry if I want to" but I have now reached the state of enlightenment. As a person with previous experience I believe that the whole ksh93 discussion was a text book example of how such things happen. There are sparcs flying and sometimes disagreement; but in the end the projects generally improve where they are changed (though there are obviously counter examples such as how "pcfs" deals with file name case) "It may be your project, but it is NOT your OpenSolaris/Solaris". Casper From joshhurst at gmail.com Wed Oct 18 08:39:51 2006 From: joshhurst at gmail.com (Josh Hurst) Date: Wed, 18 Oct 2006 17:39:51 +0200 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> References: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> Message-ID: On 10/18/06, Casper.Dik at sun.com wrote: > > >On 10/18/06, Josh Hurst wrote: > >> On 10/18/06, James Carlson wrote: > >> > Josh Hurst writes: > >> > > On 10/17/06, maoyi wrote: > >> > > > I want to post my 4over6 code.How? > >> > > > > >> > > > 4over6 is a project that makes ipv4 packets go through ipv6 network. > >> > > I suggest that you wait until the ksh93 integration project is done > >> > > with it's work. The project suffers from excessive bureaucracy > >> > > overhead and in my opinion a discussion should be started to > >> > > streamline the project integration process. The current process has > >> > > become unbearable and this needs to be corrected before new projects > >> > > are launched > >> > > >> > Your fear-mongering is completely misplaced just plain unacceptable. > >> Please understand that I am trying to start a generalised discussion > >> whether all this bureaucracy is REQUIRED. No other Open Source project > >> requires this kind of excessive bureaucracy overhead. > >> > >> I do not feel very well that I have to kickstart this discussion but I > >> think the excessive delays in the ksh93 integration project are no > >> longer bearable and require at least a look at the general process. > >> > >> I'd like to start a constructive discussion about the problems because > >> a reform is overdue > >I've added the CAB discussion list to the CC: for comments. IMO the > >problem should be discussed before any new projects are started. We > >need a new, streamlined process which makes sure that projects are no > >longer trashed with excessive bureaucracy > > > This is assuming that there is a problem; three CAB members did take > part in the development process discussions (Al, Rich and me) and > I don't think we so fault in it. Of course, we're all Solaris/Sun > fanboys and believe that the way in which we architect and develop > is "The One True Way(TM)". Is there any other Open Source project which has similar bureaucracy requirements? The integration of ksh93 is an item which the average Linux distribution does in days or weeks while OpenSolaris.org needs 8 months to present another list of bureaucracy items which still need to be finished. I think it is excessive to the point of mental injury that such a simple project gets delayed by a factor of 40. Applying the same bureaucracy to other projects will lead to the destruction of OpenSolaris.org. I predict that either the number of contributors will cease or a full fork of OpenSolaris.org (the whole organisation, not the Solaris kernel) may appear, drawing all attention and contributors to the new project -- Josh From Frank.Hofmann at Sun.COM Wed Oct 18 09:41:38 2006 From: Frank.Hofmann at Sun.COM (Frank Hofmann) Date: Wed, 18 Oct 2006 16:41:38 +0000 (GMT) Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> References: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> Message-ID: On Wed, 18 Oct 2006, Casper.Dik at Sun.COM wrote: [ ... ] > There are sparcs flying and sometimes disagreement; but in the end There's this picture in my mind now of CAB members grabbing whatever piece of Sun workstation hardware in reach and hurling it at each other... :) > the projects generally improve where they are changed (though there > are obviously counter examples such as how "pcfs" deals with file > name case) ? I know this popped up because of my last bundle of PCFS bugfixes (which actually unintentionally made PCFS behave like Windows does ... and that immediately caused a flurry of bug reports how much breakage this causes because there are expectations towards PCFS beyond the capabilities of the underlaying FAT specification ...) but I don't understand in which context this is project/CAB-related ? FrankH. > > "It may be your project, but it is NOT your OpenSolaris/Solaris". > > Casper > _______________________________________________ > opensolaris-discuss mailing list > opensolaris-discuss at opensolaris.org > From Casper.Dik at Sun.COM Wed Oct 18 08:58:37 2006 From: Casper.Dik at Sun.COM (Casper.Dik at Sun.COM) Date: Wed, 18 Oct 2006 17:58:37 +0200 Subject: [cab-discuss] Re: [osol-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: how to make a new project?) In-Reply-To: <3051065.1161183199401.JavaMail.Twebapp@oss-app2> References: <3051065.1161183199401.JavaMail.Twebapp@oss-app2> Message-ID: <200610181558.k9IFwbk6022959@vaticaan.holland.sun.com> [[Note: please use a tool which uses 72 column line breaks so people can use any tool of their choosing to read and respond to mail.]] >Bureaucracy. That's a word teetering on "understatement". Which projects did you run and which issues with the bureaucracy did you run into? >Disclaimer: I'm not a Sun employee, and have never been one. > >As I see it, there's a few shortcomings with the organization. So get involved in improving it. >1) The CAB. Where /are/ they? Maybe the CAB holds meetings and the >minutes of them are posted on s ome obscure corner of opensolaris.org. >Maybe the CAB provides regular guidance or has a way for non -CAB >members to address the CAB, but I'm overlooking the link to these items >on opensolaris.org's f ront page. Sure, there's a link to what the CAB >/is/, but there's nothing on what it has actually d one in the time >since it was formed... my point here being that the CAB from my vantage >seems non-e xistent. The OpenSolaris Constitution? Hello? I see many of >the CAB members individually post here often, but I never recall seeing >regular (or any) correspondence from the CAB as a group. [Note: I probably change virtual cap every other sentence here; I'm a CAB member, a (Open)Solaris developer, a member of the team which drafted the development process, and a Sun employee] The CAB posts in its own mailing list (cab-discuss) which I've cc'ed. All discussions are there; that's where they belong. There's the genunix.org wiki which contains the constitution-in-progress. But I don't see why all of that is so important. You don't need our guidance to build stuff. If CAB (or OGB) interference is needed then that is something which is limited to exceptional circumstances. Anything else would not scale (considering that CAB/OGB members typically have full time employment) What I see as the biggest obstacle is the fact that you still need a sponsor for putbacks and that we at Sun need to do them for you. >2) Too much Sun is in the way. It seems like you can't do anything >with OpenSolaris in terms of contribution without touching SUNW, >either by being made to interface through a Sun employee or having to >acquiesce to Sun-origined policy or SOP. Take the opensolaris.org site >as a simple example. Is there even just one non-Sun person who has a >tangible role in maintaining it and giving it direction? How about code >comitters? Is there even just one non-SUNW person who can commit code or >vote meaningfully on ARC cases? Not as far as I can tell. The fact that you cannot directly commit is a pain and something of which addressing is long overdue; the governance model explains how all the other things will come into being. You cannot give people a vote in ARC or commit access or anything else if you lack information. So we default to the existing Sun developer community and slowly grows this by inducing people from the outside into the pool of people who can have more important roles. The development processes were developed in an open manner for all to see; they are not cast in stone, mind you, but without discussion they will not change. But without trying to run the process a few times, we can't find out where the issues are. Personally, I feel that the ksh93 discussion went famously, except, perhaps, for the fact that the ksh93 team was suddenly besieged with people who they'd never heard of before once the ARC started its discussions. But once over the initial shock of "who are these people", it seemed to have gone very well. The ARC members technical arguments why certain decisions were wrong in their opinion and they swayed the ksh93 project team to make changes. I think that the majority of us agree that those changes were all for the better. And as I explained in an earlier mail, the /etc/kshrc.ksh discussion also has merit. >Participation means more than just giving Joe Random the ability to >browse and download previously unavailable source code. Participation >means giving a Joe Random the potential to have a say in direction, from >the smallest one-line code patch as a reviewer/comitter to deliberating >as a voting person on a non-trivial ARC case, and many other areas. And so they have; the fact is that "having a say" is not the same as "getting your way". I certainly hope that the ksh93 did not make changes because we /forced/ them to but because they were swayed by the technical and architectural arguments. They took part in the discussion and so did Sun folks; and the Sun folks aren't "Sun"; they don't all see the same viewpoint either. (I'm all for ksh starting in gmacs mode, others are not) >I'm not here to spout FUD or stir up emotions, as I'm a fervent user >and advocate of OpenSolaris ( the code, the concept). Just trying to >give an honest perspective. In all honesty, I can't say tha t I'm >happy with how OpenSolaris (the org) has progressed since April, 2005. >In the past I felt amb ivalent about the org, but lately I find myself >being bugged by the lack of non-SUNW inclusion regarding the >inner-workings of the org. This, I feel, is summed together is peoples' >minds and it promotes the impression (to the non-SUNW person) that >there's a hard-to-penetrate bureaucracy involved, >and even perhaps one that's tilted in favor of whatever interests SUNW >may have. Well, the ksh93 project is still prgressing nicely, as far as I can tell. What is perhaps different in Sun/Solaris culture is that project teams do not have an absolute say in what goes in and where to put it. That the discussion seems to be Sun vs "the project team" is wrong; it's just so that during the start of OpenSolaris Sun employees will weigh more heavily in discussion because they hold to the established traditions and because they are the ones in the ARCs. When OpenSolaris grows, and I do agree that the pace could have been better, this will change. But I think that a future project will find similar arguments from a non-Sun dominated ARC. >I don't want in any way disparage the efforts and (in many, many >cases, above-and-beyond) attentio n individual Sun employees have given >in terms of technical discourse, guidance and direction, but the Org as >an organization entity needs some quick-order work in terms of its >inner open(2)-ness. I don't think there's much wrong with the openess; it's progressing much like we say it is progressing, except perhaps slower. What you observe from Sun is perhaps a tip of the iceberg; and sometimes we conjure up some additional Sun employees who then play roles you haven't heard of before. It's the shock of the new and the unknown; the iceberg revealing itself to the Titanic. Before we start changing the development process, I would like to see more projects go through it; I think ksh93 is functioning as an excellent icebreaker on the unchartered waters of the OpenSolaris development process. We put much of the processes in Solaris development in place for good reasons (though there has been a tendency to put to many in place, it's now a bit simpler than before); our red tape is even worse; you are only confronted with the architectural end, we get to deal with marketing, funding, steering committees and what not. I can see how the ARCs come as a shock to an outsider; as an insider, I was shocked initially too; but when looking back over the larger project I did (Solaris privileges) I strongly believe that the ARC played a significant role in improving the quality of the project. And I see the same force at work in ksh93. (And in those cases where business decisions overruled ARC decisions, the resulting mess wasn't pretty) Casper From james.d.carlson at Sun.COM Wed Oct 18 09:07:17 2006 From: james.d.carlson at Sun.COM (James Carlson) Date: Wed, 18 Oct 2006 12:07:17 -0400 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> Message-ID: <17718.20789.183622.133534@gargle.gargle.HOWL> [Dropping opensolaris-discuss at opensolaris.org; I still don't think that is a good place for this discussion.] Josh Hurst writes: > Is there any other Open Source project which has similar bureaucracy > requirements? The integration of ksh93 is an item which the average > Linux distribution does in days or weeks while OpenSolaris.org needs 8 > months to present another list of bureaucracy items which still need > to be finished. I think it is excessive to the point of mental injury > that such a simple project gets delayed by a factor of 40. So, if Solaris discards development process in order to become essentially the same as Linux, what point is there in having Solaris? Why not just contribute to Linux instead? Isn't part of the point here that Solaris is in fact different? > Applying the same bureaucracy to other projects will lead to the > destruction of OpenSolaris.org. I don't think it's really possible to extrapolate so easily from the experience of ksh93. That's at a strange intersection of history, compatibility, standards conformance, and a host of other issues. > I predict that either the number of > contributors will cease or a full fork of OpenSolaris.org (the whole > organisation, not the Solaris kernel) may appear, drawing all > attention and contributors to the new project If that happens, I say more power to 'em. One of the good things about open software development is that if you don't think it's being run right, you can go your own direction with it instead. (For example, Quagga versus Zebra.) -- James Carlson, KISS Network Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 From Casper.Dik at Sun.COM Wed Oct 18 09:09:03 2006 From: Casper.Dik at Sun.COM (Casper.Dik at Sun.COM) Date: Wed, 18 Oct 2006 18:09:03 +0200 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> Message-ID: <200610181609.k9IG93u8025006@vaticaan.holland.sun.com> >Is there any other Open Source project which has similar bureaucracy >requirements? The integration of ksh93 is an item which the average >Linux distribution does in days or weeks while OpenSolaris.org needs 8 >months to present another list of bureaucracy items which still need >to be finished. I think it is excessive to the point of mental injury >that such a simple project gets delayed by a factor of 40. ksh93 was not added to Linux; it exists as packages in Linux distributions. Try adding something to the Linux Kernel. >Applying the same bureaucracy to other projects will lead to the >destruction of OpenSolaris.org. I predict that either the number of >contributors will cease or a full fork of OpenSolaris.org (the whole >organisation, not the Solaris kernel) may appear, drawing all >attention and contributors to the new project ksh93 is progressing nicely and the persons complaining have not yet said which projects of their own they are now delaying. As others have said, people want Solaris because of architectural stability; the price we pay is the ARC process. It is not ideal, but we haven't figured out a better way. Or can you explain why putting ksh93 in /sbin is better or what was wrong about the outcome of the ARC discussions? Casper From Casper.Dik at Sun.COM Wed Oct 18 09:11:10 2006 From: Casper.Dik at Sun.COM (Casper.Dik at Sun.COM) Date: Wed, 18 Oct 2006 18:11:10 +0200 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> Message-ID: <200610181611.k9IGBAbW025460@vaticaan.holland.sun.com> >I know this popped up because of my last bundle of PCFS bugfixes (which >actually unintentionally made PCFS behave like Windows does ... and that >immediately caused a flurry of bug reports how much breakage this causes >because there are expectations towards PCFS beyond the capabilities of the >underlaying FAT specification ...) but I don't understand in which context >this is project/CAB-related ? No, it's about a long time ago when long filenames were added to Solaris and when the original proposed "foldcase" option would only fold case of the 8.3 filenames (as windows does) but the ARC insisted on foldcase to also fold the long file names which was silly as those are case preserving. Casper From Casper.Dik at Sun.COM Wed Oct 18 09:38:26 2006 From: Casper.Dik at Sun.COM (Casper.Dik at Sun.COM) Date: Wed, 18 Oct 2006 18:38:26 +0200 Subject: [cab-discuss] Re: [osol-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: how to make a new project?) In-Reply-To: References: <3051065.1161183199401.JavaMail.Twebapp@oss-app2> <453645C1.6040400@richlowe.net> Message-ID: <200610181638.k9IGcQkb000215@vaticaan.holland.sun.com> >Normally, I would whole-heartedly agree with you, but I think there's >a difference here that requires a more visible approach on part of >the CAB. How do propose to achieve that if you don't even include cab-discuss in the mail? >Out There, many people still see OpenSolaris == Solaris == Sun, and >by way of that, direction of OpenSolaris(.org) runs the risk of A) >still being seen as "Sun's thing" and B) the CAB being a rubber stamp >entity for SUNW. The CAB does not play an active rubberstamping role; we do not approve or disapprove; we write the constitution. >This may be bit of an extreme example/analogy, but consider for a >moment the general public's perception of the relationship between >ICANN and the U.S. government. The U.S. Government took InterNIC and >spun it off as ICANN, a supposedly separate and (importantly) >impartial internationally-oriented entity. However, some actions of >ICANN's board vis a vis U.S. .gov policies have given the impression >to the general public that ICANN is not completely impartial to its >creator and its whims. > >As I said, that may be an extreme analogy, but the synopsis of the >ICANN situation rings up some similarities *in terms of perception* >here. Just s/ICANN/OpenSolaris and s/US .gov/Sun. Well, except that ICANN *IS* run by the US government and even more than before (ICANN was reorganized so that US government talks *directly* to US based ICANN employees rather than the "bloody foreigners" in the board. >Fostering the correct perception is important, regardless of how true >the current situation on the ground may actually be. As young as >OpenSolaris is, CAB has a intrinsic role in this. I don't think there's anything we can do as CAB to change the perception, if it exists other than in the mind of IBM. The only way for the perception to change if people are seen to actively contribute; we like to think that this actively contributing should be done in the spirit of Solaris: with proper architectural review and oversight by the *community*. >Changes to and the direction of OpenSolaris ultimately go through one >or a group of people, and these people are always SUNW people. Sure, >we have our discussion forums and whatnot where both Sun and non-Sun >people deliberate together, but when you get down to the brass tacks, >one has to /at least/ have a Sun ID badge to agree with and make the >actual changes have been proffered. The mere potential of an >opportunity for joe at foo.com to be one of those brass tacks people >does not exist, and if it will in the future, it isn't being clearly >spelled out and articulated. It is clear to anyone with a brain cell between them that it is *impossible* to open commit access to anyone overnight; you need to earn that right (as in other committees); just as you will need to earn other rights (such as becoming a voting ARC member) The CAB's role in this is solely to create the governance model which will cause the roles of external contributors to become clear and how these people will ascend the ranks of those with a say. Clearly this community can but be seeded with Sun's own employees at the start. Casper From al at logical-approach.com Wed Oct 18 11:45:42 2006 From: al at logical-approach.com (Al Hopper) Date: Wed, 18 Oct 2006 13:45:42 -0500 (CDT) Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: Message-ID: On Wed, 18 Oct 2006, Josh Hurst wrote: > On 10/18/06, James Carlson wrote: > > Josh Hurst writes: > > > On 10/17/06, maoyi wrote: > > > > I want to post my 4over6 code.How? > > > > > > > > 4over6 is a project that makes ipv4 packets go through ipv6 network. > > > I suggest that you wait until the ksh93 integration project is done > > > with it's work. The project suffers from excessive bureaucracy > > > overhead and in my opinion a discussion should be started to > > > streamline the project integration process. The current process has > > > become unbearable and this needs to be corrected before new projects > > > are launched > > > > Your fear-mongering is completely misplaced just plain unacceptable. > Please understand that I am trying to start a generalised discussion > whether all this bureaucracy is REQUIRED. No other Open Source project > requires this kind of excessive bureaucracy overhead. Wearing my CAB hat: What you refer to as "bureaucracy" I would refer to as applying engineering discipline to (the process of) developing rock solid, world class, highly upward compatible software. Is it necessary? Absolutely IMHO. The alternative, in terms of kernel development, is the chaos known generally as Linux. Can it (the process) be improved? Of course it can. And we (as in the CAB) and I (personally) am ready willing and able to act on any suggestions for improvement you may have to offer. > I do not feel very well that I have to kickstart this discussion but I > think the excessive delays in the ksh93 integration project are no > longer bearable and require at least a look at the general process. So your complaint is that there are "excessive delays in the ksh93 integration project" and you attribute those "no longer bearable" delays to be due to the "general process". You're not giving me very much to work with here - are you? Have you asked the developers on the ksh93 integration project why there are "excessive delays"? If so, what do they have to say? And is this something that can and should be resolved locally by the project leaders or is there really something here that the CAB needs to be involved with? > I'd like to start a constructive discussion about the problems because > a reform is overdue Sure - lets have a constructive discussion. Before we start "reform"ing anything, we must first identify the problem we are trying to resolve. We must clearly identify any road blocks, prioritize them and then deal with them. But wait - doesnt that sound like I'm already applying process to a problem arena - and you say that there is an issue with that? As regards the ksh93 integration project, I'll give you my personal assessment: - the project is going extremely well and making good progress - the project is extremely challenging, for the following reasons (not an exhaustive list): o it touches on a bunch (TM) of existing code within OpenSolaris o the well known libcmd issue o there are simply not that many talented developers who are: - intimately familiar with a complex program like a shell - who understand all the intracies of writing shell scripts - who know and understand all the related standards - who are familiar with "best practices" when writing multi-thousand line shell scripts - who are familiar with Solaris and the differences between the Solaris ksh and other, current, ksh93 shell implementations. - who understand how the shell interacts with some of the low-level kernel and system APIs. - who can engineer OpenSolaris specific changes without impacting cross platform compatibility o there are working implementations available from the project (I know because I'm running one) o its one of the first and largest community led efforts. As such, many of the developers are experiencing their first ARC experience o talented and driven developers often have strong personal technical opinions and preferences in many different technical areas o in many cases on the ksh93 project, there were/are many possible solutions to the implementation of specific features/facilities. Selecting the most appropriate one is a very difficult problem - often with no obvious solution - the technical quality of the work done so far is absolutely amazing and first rate by any set of standards AFAIK I don't care if it takes a year for the project to complete - because I'd rather it be done "right" than done "fast". We've been awaiting a ksh93 for Solaris for a long, long time and I can see the "light at the end of the tunnel" already! > > Please take it elsewhere. > Fine. Which list would you suggest? CAB? If there is some matter that can't be resolved at the Project/Community level, then the CAB/OGB is the place to take it. Regards, Al Hopper Logical Approach Inc, Plano, TX. al at logical-approach.com Voice: 972.379.2133 Fax: 972.379.2134 Timezone: US CDT OpenSolaris.Org Community Advisory Board (CAB) Member - Apr 2005 OpenSolaris Governing Board (OGB) Member - Feb 2006 From Joerg.Schilling at fokus.fraunhofer.de Wed Oct 18 08:56:22 2006 From: Joerg.Schilling at fokus.fraunhofer.de (Joerg Schilling) Date: Wed, 18 Oct 2006 17:56:22 +0200 Subject: [cab-discuss] Re: Reforming the OpenSolaris project integration process (was:Re: [osol-discuss] how to make a new project?) In-Reply-To: References: <200610181508.k9IF8kHm012877@vaticaan.holland.sun.com> Message-ID: <45364ea6.78MMuo9Ata4p/8BR%Joerg.Schilling@fokus.fraunhofer.de> Frank Hofmann wrote: > I know this popped up because of my last bundle of PCFS bugfixes (which > actually unintentionally made PCFS behave like Windows does ... and that > immediately caused a flurry of bug reports how much breakage this causes > because there are expectations towards PCFS beyond the capabilities of the > underlaying FAT specification ...) but I don't understand in which context The problem with MS-WIN is that it does not seem to create a vfat dir entry in case a name is something like "README". Solaris was the only OS that handled pcfs in a useful way with resprct to character cases..... J?rg -- EMail:joerg at schily.isdn.cs.tu-berlin.de (home) J?rg Schilling D-13353 Berlin js at cs.tu-berlin.de (uni) schilling at fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily From al at logical-approach.com Fri Oct 27 12:58:47 2006 From: al at logical-approach.com (Al Hopper) Date: Fri, 27 Oct 2006 14:58:47 -0500 (CDT) Subject: [cab-discuss] phone conference call Nov 1st noon Pacific Message-ID: I'd like to schedule a CAB call for Wed Nov 1st at noon Pacific. Items to be discussed: a) finish up any/all outstanding issues for the Governance b) define a schedule and time-line to: - get the Governance voted on/ratified by the community - implement elections for the next OpenSolaris OGB. We need to "push" to get both these items completed. RSVP only if you can't make the meeting. 1) "homework": review the latest Governance Draft 2) bring up any issues still on your mind for discussion on the cab-discuss list Regards, Al Hopper Logical Approach Inc, Plano, TX. al at logical-approach.com Voice: 972.379.2133 Fax: 972.379.2134 Timezone: US CDT OpenSolaris.Org Community Advisory Board (CAB) Member - Apr 2005 OpenSolaris Governing Board (OGB) Member - Feb 2006