Life is so good, it gets better every day

It’s been a very difficult year for most of the OGB members. We came to the table having ratified the current constitution, trying to figure out where we fit in. While I think the OGB made some good positive progress in some areas, it also feels like we lost the ability to talk with our community and pro-actively work on a plan for growth and prosperity for our community. It would be so, so easy to walk away after a year…

But I’m not. Consistency and context of past discussions is hugely important for subsequent boards, and so far in the nomination process, Alan and I are it. I’ve always said that I’ve wanted to stay on at Sun because I think I can still add value. When that day comes when I can’t add that value, and there are other better experienced people around, I’ll leave. The OGB is no exception to that (though the current constitution states there’s a maximum of a 3 year period).

So I’m in. I’d probably give myself a C+/B- on last years performance, and I’m interested to improve that this year - on retrospect, being Secretary sucked a great amount of time and I’d like to move away from that role so I can focus on the discussion rather than trying to take notes.

There’s a couple of things I’d like to see happen next year -

• At least one face-to-face OGB meeting
• At least one face-to-face Sun/OGB meeting (and an option for other meetings with other corporate members - ideally we need to get towards an Advisory board )
• Defined focus areas for each of the OGB members, outside their normal duties
• Working group to start re-assessing the current constitution and figuring out what our community has grown into, and what changes to the constitution are needed to reflect that

Bio: Glynn Foster. Almost 30. Joined Sun in 2000, working on the GNOME project. Past GNOME Foundation Board Director, and current OGB member. Joined the Project Indiana team in 2007 and looking forward to focusing energies to product a first release OpenSolaris OS later this year.

Time is flying. They always say the older you get, the quicker it goes, and that seems true for the latest release of the OpenSolaris Developer Preview, codenamed Project Indiana. The announce mail pretty much covers most of the changes since the last release, and another incremental step towards really changing the delivery model of software for Sun has been made.

Shrinking down to a single CD image has proved massively useful for me as a remote worker, and it’s given me the flexibility of testing several ISOs on the run up to the release, without hurting my broadband plan too much - I can’t help but think that it will give 1000’s of people in developing countries with poor network infrastructure an opportunity to try it out. While the application availability on pkg.opensolaris.org is still poor, the introduction of OpenOffice fills the gap for pretty much all my needs in my day job. I can now install the packages I care about, and my disk feels lighter. Thank you to everyone who’s worked on this over the last couple of months - your patience and dedication are appreciated. Thank you to everyone who have downloaded and installed it, and more importantly, given us feedback.

But controversy continues to be the compromise for that progress.

Stephen’s two blog posts, here and here nail the issues for anyone who hasn’t caught up. It’s been a roller coaster ride over the last few months, both personally, for the project and the wider community. John Plocher has been rocking on putting together a set of draft guidelines for trademark usage and branding, after the official response from Sun on the continued plan to call it OpenSolaris.

At a personal level, being on this project is massively challenging. Not only in the desire to create the best possible user experience while encouraging continued open development, but also in terms of community dynamics and finding the right line to walk between my Sun commitments and my community ones, namely the OGB. There’s no question that there has been a shift in the community, both indicated by Sun’s rightful desire to name the artifact OpenSolaris (of which I agree with), and the interesting discussion in defect.opensolaris.org around the independence of OGB members. Dalibor’s “Finishing governance before finishing bootstrapping is a bad idea” quote highlights one of the main concerns I’ve had from the start - you can’t just switch to a self governing community overnight, you grow into it. Nor can you expect to apply a model that works in one community to another. We are all different. OpenSolaris, comically, is no exception.

So where next for the project? I’m hopeful that it will turn out just fine, perhaps naively so, but I can see people trying out the developer preview and realizing that it’s not too bad. Most of all, we need to execute in a regular and predictable fashion, as a community in as transparent an environment as possible. United. Not just Sun, but everyone.

Just got to FOSS.in this morning to join in the fun during the OpenSolaris and GNOME project days there. Everyone’s pumped up about the conference and you can hear the din of people’s horns acknowledging that right across town. It promises to be a pretty excellent conference. I’m giving a talk this morning on The Secret diary of an OpenSolaris Hacker, aged 13 and 3/4 at 10am. I’ll be stepping through some of the development processes involved in contributing to the project, and helping explain how you can raise your profile in the community based on your contributions. Come join us!

One of the hotly debated issues since we launched last week was around the addition of /usr/gnu/bin to the default path, along with the addition of bash as the default user shell. I’ve been a bash user for a pretty long time, and never really found that it got in the way much (especially compared to the older generation shells). I’ve never really had much of a reason to change.

This week I’ve decided to make ksh93 my default shell, downloaded and installed from Roland’s excellent snapshots. While I use the command-line pretty much every day, I certainly don’t use it extensively for scripting, and I probably mostly care about tab completion, and clearing my terminal window.

For tab completion, ksh93 lists the options by number and details the path, allowing you to choose one by a simple ‘N<tab>’ key combination -

gman@rampage:~$ z
 1) /usr/bin/zipgrep                  10) /usr/bin/zcat
 2) /usr/bin/zsh-4.3.4                11) /usr/bin/zipcloak
 3) /usr/bin/zipsplit                 12) /usr/sbin/zlogin
 4) /usr/bin/zipnote                  13) /usr/sbin/zoneadm
 5) /usr/bin/zip                      14) /usr/sbin/zpool
 6) /usr/bin/zsh                      15) /usr/sbin/zdump
 7) /usr/bin/zipinfo                  16) /usr/sbin/zic
  /usr/bin/zenity                   17) /usr/sbin/zonecfg
 9) /usr/bin/zonename                 18) /usr/sbin/zfs

which feels pretty useful, though less useful when you type ‘gnome <tab>’ since the output is now a single column. With bash you get a similar single column output, but it is piped through more to avoid having to scroll up later. Tab completion also seems a little awkward with ksh93 if you get part of the path wrong, since it seems to add 4 spaces after you tab making it look as if the completion succeeded.

For clearing a screen, ksh93 at the moment seems a lot more irritating. The key combination is ‘<esc><ctrl>l’, whereas bash is just ‘<ctrl>l’. I’m sure the ksh93 one is doing something more, but well, gets in the way for my use.

I’ll continue for the rest of the week and see how it all works out. I do like the shnote, shtinyurl, and shtwitter scripts in /usr/demo/ksh/fun though!

Today we released the first milestone for Project Indiana, the OpenSolaris Developer Preview. Before you even read the rest of this blog entry, start your download. The locals of Guam (the last known inhabitable place in the planet to see the final hours of October pass by) are now celebrating, hopefully with some strong alcohol.

I’d like to shout out some “thank yous” right off the bat. Thanks to everyone involved in the various projects that made up this distribution, not just those on opensolaris.org but the wider free and open source community - you guys are my heros, and I strongly value your continued commitment to freedom. I’d also like to thank those projects a little closer to home that we’ve focused on for this first release, caiman, ips, and modernization (and those behind the scenes herding them - Bill, Dan, Kelly & Bonnie). They’ve survived network outages, fires *and* earthquakes to get this out the door on time - awesome! At a personal level I’d like to particularly thank David, Stephen, Dave, Danek, Sanjay and Bart for their continued patience in answering my many questions. And finally, I’d like to thank my wonderful, wonderful team mates Sara, Patrick, Jesse, Derek, Jim, Terri, and most of all Ian. We got there, woo!

Enough of the oscars, show me the software!

The developer preview is only x86 at this time (for sheer practical reasons of wanting to get something out of the door), and should run on a minimum memory requirement of 512Mb on the metal, but also in VMWare. You’ll also notice when you start playing with it -

• It is a single CD download, so much of the software you’d expect to see in Solaris Express is not there, some of which will be available from a network package repository
• It’s built on Nevada b75a
• It is also a LiveCD, allowing you to try before you install on to your disk
• Contains the latest bits of the new Caiman installer, with a significantly improved user experience
• ZFS as default filesystem - NO WAY! WAY
• IPS as the underlying network based package management system (though SVR4 packaging is still available)
• /usr/gnu/bin has been added to the default path
• bash is the default shell
• GNOME 2.20 goodies

This first release is a prototype - some indication that we really are serious about putting this together. It has come out of the proverbial sausage factory with relatively little testing and could contain bugs that could lead to panics, data corruption or other similarly uncomfortable situations. You should probably not run this in your data center.

Download it, try it, but most importantly, tell us about it! As always, we need your help - join us on indiana-discuss (or one of the other specific project aliases). We’re not done, we’re just getting started.

I’d just like to express a big Thank you to all the people who came to Santa Cruz this weekend for the OpenSolaris Developer Summit. I had an excellent time meeting everyone from right across the community, and I hope those that attended had a bunch of fun too. Hope to see the numbers grow when we do it again in 6 months time.

I’d also, of course, like to thank the people in my team for rocking so hard, especially Jesse. And thanks to Sun too (Dan and Marc) for being willing to fund the event. I’m super pumped for the developer preview release at the end of the month, and charged to make the big push for a first release in Spring 2008.

Some people doubted the summit would be successful. I say ‘HAH!’ Let me welcome the new OpenSolaris mascot, Paxton the flying pig. Awesome!

With a few weeks to go to the OpenSolaris Summit, and hopefully not much longer before we get a first preview release of Indiana, I figured it would be a good time to write an update of where the project is and the great progress that has been made to date.

The Caiman folks have been tirelessly rocking away in the background making some pretty incredible progress. Dave recently posted an updated package list for the Slim Install project, which now has a prototype successfully booting at just over 500MB in size. Considering that they started from Solaris Express and 6 CDs worth of contents, this is excellent progress that Sanjay and the team have made and represents a very solid foundation to tweak over the coming months. David is doing a similarly awesome job identifying lists of components for a first pass at an eventual network repository. Jan posted a first draft of the target instantiation spec for Slim Install, with a focus on ZFS being the root file system initially.

Karen recently announced the source code repository of the Distribution Constructor project, based on much of the work that Dave has done with the Live Media project, and Moinak’s (and team) work on BeleniX. It’s pretty nice to see this all coming together, and I very much hope that it will be similar to Revisor in functionality. You can check out the current scripts here -

  hg clone ssh://anon@hg.opensolaris.org/hg/caiman/distro_constructor

Ethan is starting to ramp up the Snap Upgrade project, with his strawman proposal for management of the boot environment, and has outlined a schedule for the work. Awesome!

And of course work is continuing on the rest of the install infrastructure after the successful integration of the Dwarf Caiman thanks to Sarah and a heap of others (William, Sundar, Niall, Matt, Jedy, Frank) - despite the occasional hiccup from the cheap seats of the audience (no offense Eric!)

Stephen and Danek have proposed their image packaging system project to the Install community, and got the all clear to set up a webpage, mailing list and Mercurial repository. Hopefully we'll see some good things soon on this front - until then, the Indiana package management status page has more details, including links to various related blog posts.

Alan has been rocking on providing an open X stack with the FOX project, and the work of Moinak and Martin. Not only that, he's had time to work on integrating libXcursor so we can finally all have whizzy animated cursors! Laca has been tirelessly building the desktop stack with the new X libraries, and with the release of GNOME 2.20 this week, we're in exceptionally good shape.

Great work is still being done by John on the i18n emancipation and Jason on an unencumbered libdisasm for SPARC. All of which looks good for being able to provide a freely re-distributable download (and torrent!).

There are other projects continuing to make significant strides too numerous to mention here (xVM guys, congrats!). Check them out, they could use your help.

And so our attention turns to next month's summit, as Ben points out. Brian has been doing a great job pulling some of the proposed topics together, so we're all coming prepared. There's 65 people currently signed up for the two day event (13th and 14th October), and it will be an excellent opportunity to meet the people behind the project and brainstorm of the coming months.

You can be 66!

Update:Okay, Simon grabbed 66. You can be 67! You get the idea...

The Summer of Code for 2007 has finished. All in all it was a pretty successful summer, although only 2 of the 4 students completed the program to the end. John did some excellent work on re-writing the encumbered internationalization code, and looks in great shape for integration in the near future. Tom did a fantastic job with porting Mlucas to the Solaris platform, and is now record holding on OpenSolaris/Sparc64 VI. I think everyone involved in the program, both mentors and students, learned heaps and I hope it was valuable to them. From an OpenSolaris point of view, we still have a bit of work to do both in identifying well defined projects with well planned steps to completion for next years participation, and continuing to reduce the bar for new people getting involved. Thanks heaps to Google for selecting us and to everyone involved this year - it continues, in my opinion, to be a very important indicator for where we are in terms of community building, and I’m looking forward to Rob Giltrap taking over the administration reigns for next year.

As our first prototype releases draws nearer, it’s time to start thinking about the branding experience for Indiana, specifically from when they put the LiveCD into their CDROM drives and launch the desktop to installing it on their disk. From my perspective, it’s an incredible opportunity to promote the OpenSolaris community brand, and create a sense of pride of ownership of the product (and technology) that we’ve taken part in. I’ve started a discussion on the Indiana mailing list - if you have some ideas of where we should focus, and what content we might want to highlight get involved! It would be great to be able to identify a solid list to work from for an eventual first release in Spring next year.