Thursday Aug 23, 2007

Silicon Valley OpenSolaris User's Group tonight: Solaris Desktop Update

Somehow our quick demo of compiz for the user group meeting tonight in San Jose turned into a full evening on X and the desktop in Solaris. Stuart and I whipped together some slides today, and have demos of both compiz and RandR 1.2 ready to go on a Toshiba Tecra M8 laptop with i945 graphics (which means it will indirectly be demoing DRI for i945 as well). For those who can't make it, the slides are here and call in numbers are on Alan Duboff's blog. For those who can make it, his blog also has the appropriate details on time and place.

Friday Jun 22, 2007

OpenSolaris and X: Year 2

open(2) logo

I'm a bit late this year - it's now a year and a week since I posted One Year of OpenSolaris and X, and it's been a busy time.

Unfortunately, I can't say that we've released lots more code than in our first year - there's still many client applications and libraries that need to be migrated to the X.Org versions - but we have moved the source base for what we've released from the Imake-built X11R6.9 monolithic source tree to source modules from the X11R7.2 release. With the 7.2 move, we also grew the set of servers delivered from the current X.Org sources from just a 32-bit x86 Xorg by adding 64-bit Xorg servers on both SPARC and x64, as well as the Xephyr nested server and Xvfb virtual frame buffer on both platforms.

Sun's China team integrated DRI support for Intel i845 through i945 graphics, with i965 coming soon. We've also added libraries from X.Org that we hadn't shipped before, including libXcomposite and libXinerama.

Work continues on converting the rest, and may pick up with some community members joining in. The OpenSolaris X community has been discussing a proposal to create a common source repository merging the X efforts of the Solaris Express, Belenix, and marTux distributions. Martin Bochnig has been working for a while on marTux, an OpenSolaris distro for SPARC - and the only one with open source SPARC Xorg drivers he's put a lot of effort into making work well on OpenSolaris. And in April, Moinak Ghosh announced he'd taken our code drops and packaging and filled in the rest of the X11R7.2 sources. By merging these, we hope to keep all distributions better aligned and make it easier for us to work together on the missing pieces.

What else are we working on? Well, like last year, the OpenSolaris suspend-and-resume team is still hard at work making their code as rock-solid as they can, including the graphics support in Xorg. We're looking at the best way to integrate Xvnc into our code base in support of the OpenSolaris Xen work. And of course, we've been playing with the projects now known as “Compiz Fusion”, using the builds from Erwann in the Desktop group, so that maybe by the time I write next year's entry, you will have a Solaris desktop option that looks a bit like this.

Thursday Jun 14, 2007

OpenSolaris Second Anniversary

Last year at the first anniversary of the OpenSolaris launch, they held a home page graphics contest to pick a graphic for the home page. Today is the second anniversary, and I didn't see a contest this year, but I have an entry anyway, suggested by conversation yesterday on the #opensolaris channel on freenode...it's hip, modern, Web 2.0, in sync with the stuff all over the Intertubes, fully overblown meme compliant:

(with apologies in advance to pretty much the entire Internet)

I can has cheeZFS?

Wednesday Jun 13, 2007

Xorg 7.2 Solaris status update

It's been a while since I've given an update on our Xorg 7.2 state in Solaris and OpenSolaris, and I was reminded by a new patch release today. For a summary of the new features included, and some bugs in the initial Nevada builds, see the Heads Up: Xorg 7.2 in Nevada build 58 message I sent to the OpenSolaris xwin-discuss list in February.

Besides the original release in Solaris Express: Community Edition and the OpenSolaris sources, Xorg 7.2 is now also available in:

For those on Solaris 10 with an older patch release, or the Solaris 10 7/07 Beta, patch 125720-08 was released today, which fixes several issues reported by early users, including:

  • A number of problems with auto-configuration of screen resolution
  • Xephyr's Caps Lock not being unlockable (bug 6539225)
  • A memory leak every time an X client connection was closed (6533650)
  • Crashes or failures using TrueType fonts in 64-bit Xorg servers (6535518)
  • xorgconfig listing the wrong keyboard driver in generated xorg.conf files (6559079)
  • Xorg -configure not including the bitstream TrueType font module in generated xorg.conf files (6546692)

Also included in the patch are:

  • A new version 2.0.2 of the “nv” open source driver for nVidia cards which adds support for the first set of GeForce 8000 series cards, though you can also get the updated nvidia closed-source/accelerated driver to support them as well. (A slightly older version of the nVidia closed driver is included in Solaris 10 7/07 already.)
  • Security fixes for two reported vulnerabilities: X Render Trapezoid divide-by-zero and XC-MISC memory corruption.
  • An update to the Radeon driver to check if you've plugged in an external monitor on your laptop (at least on Ferrari 3400's) when you run xrandr, and if so, activate the external display port without having to restart X. (A small step towards the more general and complete solution coming with X Resize-and-Rotate 1.2.)

There's a few more fixes in progress still (the -09 rev of the Solaris 10 patch is in QA now, and there will of course be more revisions in the future), including:

  • Fixing glxgears and glxinfo linking so they run (6560568)
  • Allowing Xorg to start with an old xorg.conf listing the "Keyboard" driver (capitalized) (6560332 - if you have this problem now, just change the driver name to "kbd" in your xorg.conf)
  • Adding the VMWare mouse driver, vmmouse_drv.so (6559114)
  • Fixing Xorg -configure in the 64-bit Xorg (6556115)
  • Making Xv playback work again in the ATI Radeon driver (6564910)
  • Making xorgcfg work again ( 6563653)

but the current state should be fairly usable by most people as it is today.

For those using Solaris Express/Nevada releases, or OpenSolaris sources, you can see which build the above changes went in at the X ChangeLogs page. (Solaris Express Developer Edition 5/07 corresponds to build 64 in that list.)

Removing XInitThreads from Totem

Bastien, you're not alone. If the JDS team complained to even Sun's X11 team about bugs after they applied that Totem patch, we'd tell them they're on their own and we'd never support an application that called X from multiple threads without calling XInitThreads first.

The Solaris Xlib is thread safe (though probably with a few bugs in corner cases still) based on the same X11R6 multi-thread code as everyone else, though we're currently working to reconcile the fixes from our fork with the current X.Org code as part of our ongoing project to un-fork Solaris X11 and bring it back to the X.Org code base. We're also looking at XCB in the future for a cleaner thread-safety model as well.

Update: As noted by trisk in the comments, this seems to have been caused by known Solaris libXi bug 6551484 - the engineer who had removed XInitThreads() from totem tried the test binary produced by the engineer working on the libXi bug and found it solved the hang problem she was seeing.

Monday Jun 11, 2007

The Irregular X11 DTrace Companion: Issue 2

In response to a Solaris 10 7/07 beta tester question last week, I updated the Xserver provider for DTrace web page to point out that besides the original method of patching the Xorg sources and rebuilding, pre-built binaries are now included directly in Solaris...

Getting an X server with DTrace probes built in

The Xserver dtrace providers are now included in the regular X server binaries (including Xorg, Xsun, Xvfb, Xephyr, Xnest, & Xprt) in recent Solaris releases, including:

For those who still want to build from source, the provider is integrated into the X.Org git master repository for the Xserver 1.4 release, planned to be released in August 2007 with X11R7.3, and into the OpenSolaris X code base for Nevada builds 53 and later.

Sunday Jun 10, 2007

Kinda like microblogging meets slow-mo IRC...

I was trying to explain twitter to my wife and the best I could come up with is “kinda like microblogging meets slow-mo IRC,” which didn't make any sense to her. After a bit more trying, “like a web forum with short posts” made more sense to her, but my personal dislike of web forums kept me from thinking of them like that. Fortunately more prolific and more widely read bloggers than I have spent a lot of time describing twitter and it's impact, so you can go read them for a better description.

Twitter is starting to gain popularity at Sun too, both as a new communications method to explore, and as a Solaris-hosted Web 2.0 service. They even have recently been a example of using DTrace to improve the performance of their Ruby-on-Rails platform.

So now you can also find me at http://twitter.com/alanc . I can't promise anything profound or in-depth in 140 characters, but it should get updated a bit more often than this long-form blog.

Thursday May 17, 2007

The Irregular X11 DTrace Companion: Issue 1

All the cool kids are doing unscheduled update newsletters now, and I had a collection of X11 Dtrace things to post about piling up, so I figured I'd give it a whirl. (BTW, Nouveau guys - I'd read yours more if you had an RSS feed - having to go check the wiki on an irregular basis doesn't work so well.)

Doing newsletters on a regular basis is hard, as Glynn has learned, so don't expect updates too often...

Sample script for tracing events

Jeremy asked recently for an example DTrace script using the event probes in the Xserver DTrace provider. I had lost the ones I wrote when developing/testing that probe, but dug one out of my e-mail that was written by DTrace Master Guru Bryan Cantrill and posted it as ButtonPress-events.d.

He wrote this script when he'd upgraded to Nevada build 58 and found shift-click was no longer working. Using the script he was able to show that the Xserver was sending ButtonPress events without the ShiftMask set. I wrote a corresponding script to decode the VUID events the X server was getting from the kernel mouse and keyboard drivers to determine the keyboard wasn't giving us the shift key pressed events until another key was pressed, which wasn't happening for shift-clicks with the mouse. That script can be seen in the bug report I filed with the kernel keyboard team as Solaris bug 6526932.

Using DTrace to win a race

A bug was filed with X.Org recently stating “iceauth can dump a core in auth_initialize() if a signal is caught before iceauth_filename has been malloced.” As I was fixing bugs in iceauth anyway, I figured I'd look at it to see if I could roll in a fix to the new iceauth release I'd be making for those. Fortunately, iceauth is a small program, and I could see where I thought the race condition lied between the main code and the signal handler. Testing any race condition has always been a pain in the past, since if you don't get the timing just right you don't know if you really fixed it, or just changed the timing enough to not hit it.

So I pulled up the DTrace manual and DTrace wiki and created a one-liner to force the race condition by sending the signal when I knew it was in the right spot:

 # dtrace -w -c ./iceauth -n 'pid$target::IceLockAuthFile:return { raise(1); }'

For those who haven't memorized the DTrace manual, this can be broken down as:

  • -w - allows “destructive” actions - i.e. those that change the running of the program, and don't just monitor it - in this case, allows sending a signal to the process
  • -c ./iceauth - specifies the command to run with this script, and puts the process id of the command in $target in the script
  • pid$target::IceLockAuthFile:return - sets the probe we're looking for to be the return from any call to the function IceLockAuthFile in the process matching the pid $target
  • { raise(1); } - sets the action to be run when that probe is hit to send signal 1 (aka SIGHUP) to the target process

Ran that and a few seconds later had a nice core file to verify where the race was. Turned out I was right about where the signal had to come in for the race to cause a crash, but was slightly off about the point in the signal handler where it was crashing. The fix was easy, and thanks to DTrace, easy to verify, and is now in X.Org's git repository.

DTrace Wiki: Topics: X

After last month's Silicon Valley OpenSolaris User Group meeting, Brendan asked if I'd work on the “X” topic on the DTrace Wiki. I agreed to, and have put up a small placeholder while I figure out what to flesh it out with - if you have any suggestions or requests for what you'ld like to see in a short introductory article to using DTrace to probe X, especially using, but not limited to, the Xserver DTrace provider, send me e-mail or leave a comment here.

Sunday Apr 29, 2007

3 years and still blogging...

Other Sun bloggers have been posting since Friday about the third anniversary of blogs.sun.com. Dilbert even had a very timely series of comics.

Today is actually the third anniversary of my first public blog post (though as you can see, I later migrated some earlier posts from my now abandoned internal blog to it), as I was traveling to the first X.Org Developer's Conference on the day of the initial launch, but signed up once I was there (back when signing up was simply e-mailing Will Snow to ask for an account), and started posting from there.

In my first post, I worried I wouldn't be able to find enough things to blog about - now I know I have too much to blog about, and time to write it all up is the limit - I was posting the changes in each Solaris Express release for a while, and more recently, the X ChangeLogs for each Solaris Nevada build, but both fell behind after a few months. I've got a list of links to other blogs and interesting sites that I want to post that keeps growing longer and longer as I fail to take enough time to post them here. I wasn't sure I would even get this post done during the b.s.c birthday celebration.

As some of the other bloggers mentioned, the effect has gone beyond just providing a way for any employee to post - the company's culture has changed. No longer do we complain about hearing about new announcements first in the press - blogs are now considered a crucial part of many announcement plans, especially around OpenSolaris, and bloggers are briefed in advance. I joked with Ben Rockwood at the SVOSUG last week about how much he's been yanking Sun's chain lately - how long ago would it be unimaginable that the CEO, VP's and Chief Architect get personally involved when a sysadmin at a startup complained about issues with Sun?

So much change in 3 short years - who knows where the next years will lead us?

Monday Apr 23, 2007

Compiz for Solaris

Erwann has posted Compiz 0.5.0 packages for Solaris Nevada/Express on x86 today - this is the latest step of several months of work to make this happen.

We've been working on this for several reasons. Our accessibility architect became very enthused after seeing Beryl's screen magnification support and talking to the Beryl developers at last fall's Ubuntu Developer's Summit. Our (now-former) director of Solaris Open Source (including OpenSolaris and the Solaris Desktop & X teams) became a fan after seeing the demos at the X.Org Developer's Conference Sun hosted in February. Even the Solaris CTO has jumped on the bandwagon for making the Solaris Express Developer's Edition look more modern.

We were dreading the tough choice of having to pick Beryl or Compiz - at first glance, compiz looked more stable, but beryl had more plugins/effects and the accessibility support we were interested in. Fortunately, the compiz and beryl communities solved this problem by re-uniting their efforts.

There's still work to do before this is seamless though. Compiz requires newer versions of various X components than we've had in Solaris - even February's Solaris Express Developer Edition is too old for this. We already upgraded from the X11R6.9 Xorg server to Xorg server 1.2 (from X11R7.2) in Nevada build 58 (the February SXDE was based on Nevada build 55b). I posted last week to desktop-discuss a list of where we are on the X library updates and additions needed for compiz.

Will we integrate this directly into Solaris in the future? We don't know yet, but it's definitely an option being considered. Until then, if you have nvidia graphics on Solaris x86, try it out! (Intel graphics will be an option as well soon, once the Solaris DRI team brings the Intel DRI drivers up to the level required by the X11R7.2 Xorg. We may even be able to get it running on SPARC in the not-to-distant future, once the current project to port XVR-2500 support to Xorg gets to the OpenGL stage - right now, it's mainly got the 2D portions of the driver running.)

Monday Apr 16, 2007

X11 API's for querying multi-head screen layout

I should probably stick this in the X.Org wiki or OpenSolaris/GenUnix wiki, but I'm being lazy and just summarizing to here now from an e-mail thread last week on one of the Sun internal lists...

There are now 4 sets of API's to query the layout of display screens in a logical X screen when using a setup such as Xinerama, TwinView, or MergedFB:

  1. the original X11R6.4 XPanoramiX\* API
  2. Sun's Xinerama API
  3. XFree86's Xinerama API
  4. Xrandr 1.2

The first is pretty much universally available, but also universally deprecated. The functions for it are defined in libXinerama on Linux, libXext on Solaris, but I don't think we've ever shipped the header for it on Solaris, since it was deprecated long ago.

The second Sun created and proposed to the old X.Org industry consortium to become the standard, but they did not adopt it, crafted a replacement in committee that was universally incompatible with everything and never adopted that either. It's what the <X11/extensions/xinerama.h> header in Solaris represents, and was never documented, but you can find a description of the API in PSARC 2000/036. The functions from it are found in Solaris libXext, but nowhere else in the world that I know of.

The third, aka libXinerama, is available on most XFree86 & Xorg based systems, and finally came to Solaris in nv_62 and soon in Solaris 10 patches & the next update release. It uses the <X11/extensions/Xinerama.h> (note the capitalization compared to the other) header and is documented in the man page we created for the Solaris integration and contributed back to X.Org.

The fourth is a very recent development, having been released after X11R7.2's release in February. X.Org is currently in the process of releasing version 1.3 of the Xorg server package (release candidate 5 is out now) - this will be the first Xorg server feature release not tied to one of the whole X Window System releases. (X11R7.2 had Xorg server 1.2, X11R7.3 is expected to have Xorg server 1.4.) The xf86-video-intel driver 2.0 release candidate 4 is also out, it is the first driver to support the new Xrandr 1.2 additions, though work is in progress on other drivers, including the Radeon and nvidia drivers. We don't have a definite roadmap for including this in Solaris yet - we'll probably do it once the X.Org releases are finished and we're passed the upcoming milestone of the second Solaris Express Developer Edition release in May.

Friday Mar 23, 2007

OGB Elections: Contributors and Voters

After some discussions on #opensolaris IRC today, I downloaded the lists of contributors from poll.opensolaris.org and ran some stats. Each OpenSolaris community was asked to name both Contributors and Core Contributors over the months leading up to the election - the Core Contributors are those eligible to vote. By my count, there's 268 unique Core Contributors across the communities.

First, the breakdown of Contributors by community - “unique” refers to those in each category which are not listed as either a contributor or core contributor by any other community:

CommunityCore ContributorsContributors
TotalUniqueTotalUnique
Academic and Research 3 2 3 1
Appliances 0 0 0 0
Approachability 1 0 14 8
Architecture Process and Tools 9 5 4717
BrandZ 4 1 4 3
Chinese Users 0 0 0 0
Community Advisory Board (CAB)14 8 3 0
DTrace 5 1 0 0
Databases 0 0 0 0
Desktop 8 8 6 4
Device Drivers 0 0 0 0
Documentation 5 2 12 8
Fault Management 6 1 3 0
GNU Solaris 0 0 0 0
Games on OpenSolaris 0 0 0 0
HPC Developer 0 0 0 0
Immigrants 2 0 1 1
Installation and Packaging10 6 4 1
Internationalization and Localization 5 2 11 8
Laptop11 4 0 0
Marketing 6 0 3 1
CommunityCore ContributorsContributors
TotalUniqueTotalUnique
Modular Debugger (MDB) 5 0 0 0
NFS 3 2 0 0
Networking 8 2 15 2
OS/Net (ON)4816 8 0
Observability 0 0 0 0
OpenSolaris Printing 0 0 0 0
Performance 8 4 1 0
PowerPC 4 2 11 5
Security 8 4 4 0
Service Management Facility (smf(5)) 8 1 16 8
Solaris Volume Manager 0 0 0 0
Storage 0 0 0 0
Systems Administrators 0 0 0 0
Testing 9 5 2 1
Tools22 8 20 6
Unix File Systems (UFS) 0 0 0 0
User Groups8866 1 1
X Window System 3 3 0 0
Xen15 7 1 0
ZFS14 8 1 0
Zones 7 2 3 1

As you can see 13 communities have named no contributors, which will probably result in the new OGB taking a close look at them to determine if they are really dormant or just need leaders to step up and organize them.

I also did counts of how many contributors were recognized by multiple communities as either a contributor or core contributor - these aren't necessarily the biggest contributors to OpenSolaris, just those with the widest breadth of contributions that were recognized:

# of Communities# of peopleContributors
7 2Adam H. Leventhal, James D. Carlson
6 2Casper H.S. Dik, Mike Shapiro
5 3Daniel B. Price, Eric Schrock, Stephen Hahn
4 9Bart Smaalders, Ben Rockwood, Bill Sommerfeld, Dan McDonald, Darren Reed, Kais Belgaied, Keith M. Wesolowski, Peter Memishian, Richard Lowe
3 25
2 68
1245

If you're active in one of the communities and didn't get named as a contributor, it's too late for this election, but for future status you should ask the leaders of the community what criteria they used to choose the contributors and how you can be considered for recognition.

If you were named a core contributor and haven't voted yet - hurry up! You've got a little less than 72 hours left, and while we've just passed 100 votes cast, approval of the constitution requires 50% of the eligible voters, so we need at least 30 more (and that's assuming most people voted yes on ratification). Instructions on voting are at http://opensolaris.org/os/project/website/poll_instructions/. Two tips from debugging problems reported by a pair of voters today:

  • The machine name is poll.opensolaris.org, not polls. Four engineers on IRC could not figure out why none of us could ssh to polls.opensolaris.org and were just getting mysterious connection closed messages, until stevel piped up that we were all misspelling the hostname and hadn't noticed, because the ssh/socks5 proxies we were all using didn't actually report hostname unknown errors.
  • If you copied and pasted your ssh key from a terminal window that caused line breaks to be inserted in the middle of all the ascii hex dump portion of the key, remove them - the web form allows you to put them in and doesn't warn you that it's invalid. (Again, thanks to stevel for coming to the rescue to solve that user's problem.)

Tuesday Mar 20, 2007

How to get better ATI drivers for Xorg

Just a quick note... Michael Larabel's post on “How To Get Better ATI Linux Support” is spot on with what we've heard from ATI and from others about ATI in various conversations — if you really want to get their attention, go through their OEM's and resellers, whether that be asking for improvements to their closed source drivers or ports of them to other OS'es, or even opening more specs/sources to the open source community.

(As for the competition, the same goes for them, and Sun is an nVidia reseller, so if there's something missing from their drivers that would encourage you to buy more Sun workstations with nVidia cards, let us know, and we can work it through our channels.)

[Technorati Tags: , , , , ]

Sunday Mar 11, 2007

OGB Elections: Proper Positioning

I haven't had a chance to write full position papers like some of the other candidates, so for my positions, you'll mainly have to rely on these answers I wrote to some of the questions submitted via e-mail to the candidates:

I did vote in the poll being used to test the OpenSolaris elections system, so will share my vote there, and explain some of my choices to help further show my positions on the issues facing the community.

The poll asked (and the order of these is the one I was shown - the order is randomized for each voter to reduce bias if people just start voting "1 2 3 4..."):

QUESTION 1.1: ("Priorities") Which of the following items, presented
in a randomized order, should be prioritized by the Governing Board in
order to promote OpenSolaris and increase its developer mindshare?

1 - Deploy a public defect management system ('Defect_Mgmt')
2 - Remove inactive Communities or Projects ('Remove_Inactive')
3 - Provide an x86/AMD64-based kernel/project build facility
    ('X64_Build')
4 - Deploy a public Request To Integrate (RTI) system ('RTI')
5 - Provide a SPARC-based kernel/project build facility
    ('SPARC_Build')
6 - Create an infrastructure project to run opensolaris.org
    ('Infrastructure')
7 - Replace or remove Jive forum interface ('No_Jive')
8 - Reorganize the existing Community/Project organization
    ('Reorganize_Community')
9 - Replace opensolaris.org tools with an open source CMS or wiki
    ('New_CMS')
10 - Deploy a public wiki on opensolaris.org ('Wiki')
11 - Deploy a public code review facility on opensolaris.org
    ('Code_Review')
12 - Eliminate reliability issues with opensolaris.org web pages
    ('Reliability')

My vote was:

RESPONSE 1.1 --> 1 6 9 8 4 5 3 11 10 2 7
VERIFY:  Your ballot is:
Priorities: [Defect_Mgmt Infrastructure New_CMS Reorganize_Community
 RTI SPARC_Build X64_Build Code_Review Wiki Remove_Inactive No_Jive ]

So why these?

Deploy a public defect management system

Right now bugs.opensolaris.org is not truly a two-way communication system - for people outside Sun, it's Write Once Read Many (WORM) - they can file a bug, search for bugs, and read bugs, but not update them or have them assigned to them to work on. Additionally many of the bugs and bug fields are still restricted to internal users only, though that's a harder nut to crack due to various legal and customer privacy issues.

This means projects with external contributors are at a gross disadvantage, and has led to projects such as ksh93 using a separate bug tracking system (Steve Lau graciously set up a bugzilla for them on his system), which segregates their bugs from the rest of the community, so bugs.os.o no longer has all the relevant data. This needs to be solved before non-Sun employees can be first-class members of the development community.

Create an infrastructure project to run opensolaris.org

Right now, too many improvements are stalled because the team running opensolaris.org can't keep up with the everything they need to do to implement them. Open it up now, let the community help you, and let us all move forward.

Replace opensolaris.org tools with an open source CMS or wiki

The current infrastructure is customized and weird - Project & Community Leaders are the only ones who can edit pages in their areas, so to add someone to edit anything, even just one project subpage, they have to be made a leader, even if they're not doing any leadership. The syntax is also strange and confusing to those used to just about any other wiki-like system in the world.

Reorganize the existing Community/Project organization

This has been discussed to death elsewhere already, and a big part is just cleaning up the original set of communities created before projects were done. The election confused many people since those doing the most work (projects) are not given voting rights, while those just interested in the technology (communities) are.

Deploy a public Request To Integrate (RTI) system

This is specific to the ON family of consolidations that use the RTI system (Desktop consolidations, including X, don't - I don't know about all the others), but once the code repository is outside, this will be the biggest obstacle to letting non-Sun people commit code on their own without any internal sponsor.

Provide a SPARC-based kernel/project build facility
Provide an x86/AMD64-based kernel/project build facility

SPARC is higher priority than x86, since more people have x86 system access on their own, but commits to most OpenSolaris consolidations require you to build and test on both platforms first, and that's hard to do if you aren't working at Sun or another place with both kinds of machines around and installed with the build environment.

Deploy a public code review facility on opensolaris.org
Deploy a public wiki on opensolaris.org

For both of these, while it would be nice to have on opensolaris.org, we've got community ones already that work fine - for instance, I'd be happy just making wiki.opensolaris.org redirect to http://www.genunix.org/wiki/ and declaring that one done.

Remove inactive Communities or Projects

These clutter up crowded pages and cause confusion, but don't stop any real work from being done.

Replace or remove Jive forum interface

Again, as much as it annoys me, it's not preventing real work, and can be mostly ignored if you truly hate it. (It would be nice if it didn't break threads and cross-posting and such for people who use the mailing lists instead.)

Eliminate reliability issues with opensolaris.org web pages

As much as I was tempted to vote for this just to tweak stevel, I just haven't seen enough of a reliability issue to declare this a major problem that needs attention, so I just left it off my choices completely.

Saturday Mar 10, 2007

OGB Elections: Biography of a Candidate

As you may have seen, I am a candidate for the OpenSolaris Governing Board, and as the other candidates are doing, I'm writing a bio to give background on myself. I've previously posted a more personal summary of my background, so I'll focus here on topics I believe are more relevant to being an OGB candidate. If you have no interest in the OGB elections, or in long rambling non-linear self-promoting stories, you can just move ahead to the next entry in your RSS feed reader now.

First the basics that all the candidates are stating: As you may have guessed from the URL this is posted at, Sun Microsystems, Inc. has been my employer for the past 8 years, but I have never built my own kernel. (I have designed, implemented, and integrated my own DTrace provider though.)

While attending the University of California, Berkeley I became involved in the running of Berkeley's Open Computing Facility, a somewhat unusual student group who provided access to Unix workstations and servers in a time when that was hard to come by for a non-engineering student. When I first started volunteering there in 1991, it was the only place most students on campus could get e-mail and Usenet access. Later it became the first place most students could host web pages, and continued to evolve over time. It was run by a completely volunteer student organization, in which I served first as a member of the system administration staff (their Apollo Domain/OS workstations were the first Unix machines I got root on, and the first time I got to have the account name alanc instead of just a class account name like cs60a-ej). I later joined the OCF's Board of Directors, who set the policies for the machines, and served a semester each as Site Manager (head sysadmin) and General Manager (head of the organization).

It was also at the OCF that I first started contributing to Open Source projects. My first major contributions were to the Gopher client/server software for Unix platforms from the University of Michigan, an early precursor to the world wide web. Of course, my most notable open source work has been through my job at Sun, contributing to the X Window System software that forms the foundation of all major Unix and Linux GUIs today. I participated in the old closed-membership industry consortium X.Org Group on behalf of Sun, and contributed to their irregular public code drops of the very closed development tree, and also contributed some of our code to the XFree86 Project, including the support I wrote for IPv6 in X11. When we were looking for ways to revitalize the stagnant X.Org Group at the same time that a group of current and former XFree86 developers were looking to create a more open development model to replace XFree86, I got to be one of the developers in at the very beginning of the transformation of both groups into the X.Org Foundation and it's Xorg software releases. I was elected to the short-lived X.Org Architecture Board, and was the release manager for the X11R6.9 release and one of the leads of the modularization project that resulted in X11R7.0. Through this I've experienced first hand how much more vibrant and productive a fully open development community can be than one in which a small core team holds the leash to all access, and all changes have to be funneled through them. (OpenSolaris isn't quite as bad, as there are several dozen “request sponsors” who can check in code from community members, but we still see requests that take much longer than they should because sponsors simply get too busy and become an unnecessary bottleneck. We've got several external contributors who have clearly proven they deserve commit access, but we haven't gotten the master repositories for anything besides JDS outside the Sun firewall yet so they can exercise that.)

When I took some time off from college, the Unix administration and end-user communication skills I got at the OCF, combined with the connections I had from that community, got me a position as a contractor in Sun's technical support center (then in Mountain View, California) answering Solaris system administration questions. Since I was more familiar then with other forms of Unix, and already had been addicted to Usenet for years, I looked for help in learning the Solaris specific methods to the comp.unix.solaris Usenet newsgroup, and returned the help I got by answering the questions that I could. (And even though I haven't posted as much lately, I just noticed Google still shows me in the top ten all-time posters to comp.unix.solaris, under my alum.calberkeley.org address.) It was through comp.unix.solaris that I first participated in the Solaris community, and first came in contact with current OGB members Rich Teer and Casper Dik, and of course, ksh93 for OpenSolaris project leader Roland Mainz.

After joining Sun I also started participating in the Solaris x86 YahooGroups mailing list, sticking with it even during the dark days in which Sun had declared Solaris 9 for x86 “indefinitely delayed.” After the “Secret Six” from that group convinced Sun to restore Solaris x86, one of the projects Sun started to restore relationship with the community was monthly meetings with a “Cabal” of community leaders. When we decided to ship the Xorg server in Solaris, I was asked to come to one of these meetings to talk about our plans, and given my involvement in the community already, was invited to become a permanent part of the group. That group didn't last long, as it ended when the participants, including me, became some of the original members of the OpenSolaris Pilot Program, so I've been part of the OpenSolaris community since the very beginning. Since the public launch of OpenSolaris I've served as a leader of the OpenSolaris X Window System Community and the representative of the X Consolidation to the OpenSolaris Program Team. (The OpenSolaris.org forum software says I've posted 1,190 times to OpenSolaris forums since the launch, but I don't know how many of those are duplicates from cross posts to multiple forums.)

At Sun, I've also served for about 6 years now on the Desktop C-Team, which oversees the various Desktop consolidations (X, GNOME, Mozilla, CDE, etc.) - approving integration of new projects, setting and enforcing various policies, and coordinating joint projects with other consolidations. For the past three years I've been part of the Layered Software Architecture Review Committee, first as an intern, and now as a full member. LSARC is the ARC which reviews software above the core platform - including the desktop environments, web browsers, compilers and developer tools, and various management tools. X itself is usually straddling the boundary between LSARC and the core platform ARC (PSARC) - cases interfacing with the hardware or kernel go to PSARC (including most updates to the X servers themselves), while client libraries that are mostly used by the desktop environments go to LSARC, so that they fit into the larger picture of those desktops. As such, even though I'm part of LSARC, from time to time I submit cases to PSARC and participate in relevant PSARC reviews.

So that's the long boring story of how I got to where am I today (if you've made it this far, thanks for sticking through it, and be glad that I didn't make it even longer and more boring with recounting my experience as editor of the high school newspaper or how bad I was at memorizing my lines in high school Drama Club productions). Now that we've covered the past, I'll save the topic of the future and my positions on various OGB issues for another post...

About

Engineer working on Oracle Solaris and with the X.Org open source community.

Disclaimer

The views expressed on this blog are my own and do not necessarily reflect the views of Oracle, the X.Org Foundation, or anyone else.

See Also
Follow me on twitter

Search

Categories
Archives
« April 2015
SunMonTueWedThuFriSat
   
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
  
       
Today