Once More unto the Breach

Perfectly Reasonable Deviations From the Beaten Track

My apologies that this is a little off topic.  A new Feynman book is coming.  His daughter has edited and assembled his letters.  A small example is in the print edition of the March 2005 Discover magazine (Vol. 26, No. 3), with the promise of the book out in April 2005.  Can't wait.

• Perfectly Reasonable Deviations From The Beaten Track: The Letters Of Richard P. Feynman

Microsoft Trying to Release More Source Code

Excellent news as always.  The opportunity for Microsoft is still huge in this space.  They have the opportunity to lead in the community (and I don't mean lead the community) if only they would take it.  This is actual product code, so it's stickier to release into community than non-product code like WiX, WTL, and FlexWiki.  Go Shawn.  Go Jason.

Small editorial note on the article:  Microsoft took the C# and CLI specifications to ECMA (and the ECMA technical committee subsequently took the finished ECMA standards to ISO).  Rotor (aka the Shared Source CLI) was a working implementation of the ECMA standards on XP, FreeBSD, and Mac OS X. 

Links:

• "Microsoft to Release More Source Code?"
• Microsoft and community development
• Rotor (a shared source project)
• The ECMA specifications for C# (ECMA 334) and the CLI (ECMA 335) (freely available)

Linus, and Career Path, and the Summer of Love

An InformationWeek blog entry by Larry Greenemeier appeared during last week's OSDL Enterprise Linux Summit entitled "Open Source is Not a Career Path” covering part of a great panel discussion.  (I blogged it poorly as a set of audience questions, partial answers and notable quotes.)  The InformationWeek blog entry is missing in action at this point (I needed to reach for the Google cache.)  The storm of responses from the /. post as well as the InformationWeek post was pretty impressive.

What most impressed me was how off context the initial IW piece seemed to be.  The audience question was “What advice would the panel have on an OSS career path for a new developer?”  The initial response from Linus was “It's not a career path; it's a learning experience.”  Without putting words in the man's mouth, I read the intent of the answer (and other panelist answers) more in the light of the open source development model represents a software engineering practice, and positioning their responses in a career oriented way.  Saying open source is a career path would be like saying software engineering is a career path. 

The responses around motivation and passion were pretty normal.  Think about the advice most people give around any career choice: do what you're passionate about.  Do what you care about.  Do what you love to do.  Developers that don't care about a particular area of an OSS project (and care might mean a corporate pay check) aren't likely to show up and work on it.  The original author's glib “Only the truly passionate need apply” demonstrated a perfect lack of understanding in an otherwise accurately reported piece (if somewhat lacking in context.) Then the storm of responses.   

So many people want to turn open source into a counter-culture debate.  The press, IBM marketing, and Microsoft's norm entrepeneurs reach for the best counter-culture icon they have in the Summer of Love, and “Peace, Love, and Linux”.  Unfortunately, as an icon, it just doesn't fit.  Most of the open source developers I've known for the past 15 years are also some of the most free market oriented people I know. They also happen to be very passionate about what they do, and that's all any of us can hope for in life.

Links:

• The slashdot post for "Open Source is Not a Career Path"
• Po Bronson and "What Should I Do with My Life"
• IBM and the Summer of Love
• Where we're going

 

OSDL Summit: Wrap-up

I caught two last sessions today.  First was IBM's Gerrit Heizenga and "The Relative Maturity of OSS Projects", followed by the ever entertaining Bdale Garbee (HP) talking about "Helping Enterprise Architects Make the Transition". 

It was interesting to see the positioning differences between the two talks based on the speakers' employers.  Gerrit's talk was focused on the nuts and bolts of how to think about approaching new community projects, talking about what sorts of things might be interesting for you as an enterprise to evaluate.  His use of some dangling examples was causing a certain amount of teeth grinding with another vendor in the room.  Bdale's talk was more focused on helping architects from a philosophical position look at their needs and understand how to think about community.  Bdale's talk was good because it turned the view inside out, helping architects look through what is perceived as the "risks" of open source, and instead evaluate their world from the options, choice, and opportunities OSS provides. 

It has been a great little conference.  It seems to fall some where comfortably between the very CIO focussed feel of the Open Source Business Conference, and the developer centric community world of O'Reilly's OSCon.  For the track I've been following it was very nuts and bolts focused at the enterprise management level, providing a framework and roadmap for how to think about open source in the enterprise.  For a first time event it's been great.  (The official gate was 327.)   

Links:

• The Relative Maturity of OSS Projects
• Helping Enterprise Architects Make the Transition
• Open Source Business Conference, April 5-6, 2005, San Francisco, CA
• O'Reilly Open Source Conference, August 1-5, 2005, Portland, OR

OSDL Summit: Tim Golden on creating OSS stacks

Day 3: First talk of the day was Tim Golden from Bank of America talking about "Successful Open Source Software Stack Creation".  A good talk on his 16 steps to developing a platform stack.  A very methodical and flexible set of best practices that was well laid out.

Link:

• Successful Open Source Software Stack Creation

OSDL Summit: Brian Behlendorf on OSS Development

This was my last real session of the day.  (Okay yesterday.  It's late.  I'm a little behind.)  It was billed as "Applying OSS Practices to Corporate Software Development".  But I cannot do this justice.  When you've hung out at these events enough, you start to get the feeling in the first few minutes of any talk that you know where the talk's going and you could probably do a passable job delivering it.  This was a genuinely fresh talk.  And it was Brian so you know it was brilliantly delivered.  Rather than me trying to cover the talk, and waiting for the slides to get the official show template and published out, Brian has graciously allowed me to put his slides up here.  He does ask your forgiveness for a couple of the slides as once again he was developing a talk at 5am the day he was due to give it.  Enjoy. 

Links:

• Applying OSS Practices to Corporate Software Development
• The Slides

OSDL Summit: Linus, Andrew, Mitch and Brian

A wonderful panel:  Linus Torvalds, Andrew Morton, Mitch Kapor, and Brian Behlendorf, moderated by Stuart Cohen.  It was billed as "An Open Conversation with Linus and Other OSS Notables". 

One can never cover these things properly.  I don't touch type nearly fast enough, and hopefully someone has actually recorded this one.  But in the mean time here are some of the more interesting questions, answers, and quotes.  My apologies to the participants, if I slipped a word here or there, or incompletely covered the answers.

Things started with a discussion around the perception propagated in the press of "no one in charge" compared to a monolithic central development org. 
Brian: "Is Microsoft criticizing us for not having a single point of failure?"

On Software Patents:
Linus: Noted software patents aren't particularly useful and that hopefully the vendors are starting to realize this too.
Mitch:  Raised concerns for the number of bad patents and the failure to
effectively weed them out in the review process, rather than letting the courts
figure it out.  He used the analogy that we're building to a patent Bhopal and likened the situation to intellectual property Weapons of Mass Destruction.  [srw - my naive opinion is he's contrasting two things that aren't directly related, as in "patents bad;  OSS good", but the discussion was fast and furious at this point.]
Stuart: Supported the great first steps of IBM and Sun. 

Audience Question: What's the vision for OSS 5 years out?
Linus: "I am the anti-visionary.  Visions are dangerous."  He was concerned with looking beyond the ground in front of you.    "It's a lot more important to know where I am
today rather than dream about where I might be in the future. "
Mitch: OSS is an irresistible force.  Immovable objects will be encountered and the outcome can't be predicted.  Beyond the software space: he looks forward to what's next in the world for collaborative development processes.  He used the Wikipedia example (while acknowledging it as a controversial one) "We are at the beginning, not the end. "
Brian: His is a vision of open networks of collaboration.  He observed that it's easiest to see the collaboration play out in software and network standards, and likened OSS development versus traditional closed development to market economies versus planned economies.

Audience Question: Can the participants comment on the relation of open source and open standards?
Brian:  They are two sides of the same coin.  Support for the LSB and IETF are
fundamentally important. He also observed that an OSS de facto technology is better than a proprietary de facto technology because of the available access.
Linus:  He raised concerns that open standards are often imperfectly formed (when developed too early) and if there are only proprietary implementations, there may not be the incentive to "fix" such a poor standard.

Audience Question: When corporations participate in community, do they work on things that advantage their companies?  How does it all work?
Linus:  It has allowed companies to further their own agenda.  It is driving Linux in real ways.  You have to have motivation.  It doesn't come from the project internally but comes from the external needs and improvements.  The problem is you have to have the infrastructure to manage the resources that show up.  He believes the management infrastructure has evolved well over the 15 years as Linux has evolved.  There have been surprisingly few politics.  "The bad engineers don't matter and the good engineers don't play politics."  He doesn't ever remember seeing a company "position" their code.
Andrew:  "We just subvert them."  They become more loyal to the code than their employer.  He observed that it is essentially the "customer" sending in the code and that the things they need they write.
Brian:  You have to assume that everyone brings their own agenda.  The hobbyist can take the biggest risk in development without the fear of being fired.  If you're a community leader, you need to foster a diversity of motivations and employers and biases. You can generally see malevolent contributions coming.

Audience Question: What advice would the panel have on an OSS career path for a new developer. 
Linus: "It's not a career path.  It's a learning experience." Its the motivation question.  You need to have a motivation to the project.  The question isn't "What would you like me to do?"  It is "What do you want to do?"
Mitch:  Observed there is a quality of work environment in an OSS "job" in its peer interaction and expected quality of the work itself. 
Brian:  Tackled the question as more of the traits of an OSS developer.  To be successful you must decisive.  You must communicate clearly.  There's not much room for prima donnas.  You need to be able to defend your decisions appropriately. He believes traditional closed software companies have a higher tolerance for the "bad" traits displayed by developers, i.e. the lack of communications skills etc.

Closing remarks (what does the OSS world "need"?):
Mitch:  "From 35,000 feet, if it's not broken, don't fix it."
Brian:  Finding ways to turn users into developers would be great.  Be willing to run some experiments and take some risk.  Recognize the freedom of choice that comes with participation and use.  Find an open source developer and learn about the process.  Make it personal.  Re-humanize the process of software development.
Andrew:  Raised concerns that the process of adoption of open source is inefficient.  Companies are continually having to re-invent answers to questions of "how do we evaluate open source,  how do we join the community, how do we understand the legal implications?"  The educational process needs to be solved. 
Linus:  Co-operation is the motivation.  It is sometimes more fun to compete and great software happens in that mix.  It is great to say we want to co-operate on the desktop, but "let's not co-operate too much.  Let's leave some room for our differences."

An awesome panel.  My apologies again for not quite covering it.

Links:

• An Open Conversation with Linus and Other OSS Notables
• Comments on the Wikipedia example

OSDL Summit: Terence Sherlock (HP) on Linux Readiness

I hung out for most of the talk (Linux Readiness for Enterprise Deployment Core to Edge) in this first session of the day.  The thing that struck me was it seemed an HP take on the IBM Steve Mills LinuxWorld keynote of two years ago.  At that time, in New York City, in front of a room full of Wall Street CIOs, Steve Mills gave a deliberately boring keynote.  I don't mean that in a bad way: that was the message.  This is no longer bleeding edge technology, but is business ready.  (He's a very dynamic speaker but the talk was one example after another of replaced all these servers with these Linux servers, even IBM runs its business on Linux, and AIX will one day get replaced - but not for ten years.)  That's the flavour of this morning's session.  Several analyst based slides of data on growth, stability, and futures.  Lot's of Christensen-based observations (but shallow).  You can bet your business.  Linux will be ready for the core enterprise servers real-soon-now, but not quite yet replacing the [HP] UNIX big iron core.  A feel good about Linux in the enterprise message.

On the futures side, talk was focused on "ripping out Microsoft Office and replacing it with Linux" not happening yet.  That is a Linux focus, and to be fair this was a Linux talk.  The more relevant observation is when will people start "ripping out Microsoft Office and replacing it with OpenOffice on Windows."  I've been living on OO for a few months now and am really really happy.  And I can dump Microsoft Word format for people that need it.  And PDF.  And [readable] HTML.

Links:

• Linux Readiness for Enterprise Deployment Core to Edge