ADSK and the MapServer Foundation - tying the threads together

[Gary Lang]

Folks,

 

I want to answer questions and continue to try to learn what the
concerns are. I thought I'd sit down and write a longer mail about the
discussions I see taking place around all this. The responses seem to
range from thoughtful to paranoid. So clearly the inability to speak
freely has not been our friend. I have triedi to explain why we needed
to be quiet about it, and hopefully that adds to the understanding. 

 

It is good to be able to speak freely about all this, at last. I have
been very uncomfortable with only being able to speak to a limited set
of stakeholders, but unable to see a way around the legal issues our
legal team put in front of us before undertaking all this. Let me give
some background on how we got here. 



About two years ago I kicked off a project to rewrite MapGuide so that
it would be a first-class citizen on Linux. The current MapGuide is
Windows-centric to the core.  To me, trying to do web-mapping in 2003
that didn't support Linux was a non-starter. The team agreed and off we
went.

 

We started off using ACE and BDB as foundation technologies, used SWIG
for generating our API infrastructure, and switched to SQLite for the
geodatabase, and created the FDO framework for accessing multiple data
types through a modular architecture that is sort of like JDO but with a
geospatial slant (hence "Feature Data Object"). At the start of the
project, on instinct, I asked the engineers to please make their code as
readable and as easy to change as possible, as if someone from outside
the company would read it some day.

 

We really valued the open source components we were using and started to
wonder how to give something back to the open source world. About a year
ago, we visited a friend of mine who was the CEO at Ximian, which had
been recently acquired by Novell. We spent a lot of time with Nat and
Miguel who encouraged us to think hard about open source. We were
thinking about some minor products. They encouraged us to think about
Tux after hearing about it. We listened carefully and took lots of
notes. 

 

A little less than a year ago, I began to believe that, as we were
selling more applications the web server components, maybe our users
would be better served by an open source process for the Map server
tier. It turned out that DMSolutions and some of our people in Ottawa
worked at the same company a few years ago, and we got together through
one of these mutual ex-Tydav people. I decided to disclose them on our
idea to see if it made any sense, and initially to try to convince them
to maybe switch. Imagine the headlines - "Autodesk Goes Open Source!
DMSolutions Ditches MapServer for MapGuide!." This was a dumb idea, and
a non-starter for DMS. 


Instead of convincing them to switch, they convinced us to work with the
MapServer community in a way that was non-competitive. To explore this
further, we asked some of the recommended folks to sign an NDA so that
we could tell them about Tux and about our crazy idea to be the first
vendor in this space to go open source with their product, and a new one
at that.


To answer one of your questions: there is no perceived benefit to this
being a secret. In fact, it's not good from any perspective including
ours. There simply was no other way for us to explore this idea with the
MapServer community at large. Let me explain.

In Ed's email, he referenced the following comment: "And you replied,
"You do it in an open, inclusive way that opens the discussion to all
stakeholders, not just a self-selected few.  You don't ask us to sign
non-disclosure agreements and deliberately exclude the majority of the
community."

I would ask folks to please go back and read Dave's email with respect
to Autodesk being a public company. Unfortunately, we do not have the
freedom to speculate in public or hand out our source code without some
sort of framework like the foundation or an NDA. So we had two choices:



1)     We could have gone it alone and been public about that from day
one. However, that gave us no options for stakeholder consultation and
meant we would have to try to bridge the gap with MapServer after
effectively attacking it.



2)     Or, from day one, we could ask those who represent the widest
number of MapServer stakeholders - i.e. the steering committee - if we
could have a preliminary conversation about what to do. 



Those were really our only two choices. The bottom line is that a public
company can't speculate on a mailing list about what it should or should
not do. If we disagree on this point, let's send our lawyers off to
argue about it, but I am not in a position to disagree with ours in any
event.

Now that the cat's out of the bag, I can say that we are much happier
approaching open source this way rather than doing it ourselves. We hope
others will feel the same once they understand the two options we had.
Maybe we could have been more creative and found a third way. As a model
for future participants or even to help me in any future open source
work I'll do (and I'll do more in my career, I am certain), any
suggestions would be appreciated.

Some things you will see as the weeks progress with respect to
Autodesk's involvement:

a) We will participate in establishing the foundation, which, contrary
to what has been said here, is already a legal entity. However, we will
not drive the foundation. It would undermine our goals to do this. 

b) We will give our code - about 60 man-years of sweat and blood by
people who love their work--to the foundation. This is a non-trivial
thing. Do the math on the cost and think about projects you've worked on
that take 2 years of hard work. We are contributing something too.

c) we will be a minority vote in any foundation configuration. I am
limiting Autodesk's participation to two people - myself and one other
person - both engineers. I expect the MapServer membership to be some
multiple of that. As it is, the current discussion and decision process
observes the ratio you see on the open letter signature list - 1 to 10,
Autodesk to MapServer. We will never have control of the MapServer
Foundation's board.



d) we will contribute code to MapServer as well as continue to work on
the code we 
are contributing which I'll refer to as MSE for short. We will give away
our FDO framework for ArcSDE, SHP, ODBC, MySQL, etc. We will make this
work with MapServer.

Finally, I hope people can continue to be respectful of the signers of
the open letter. They are nobody's fools and nobody's tools and will
simply use this opportunity to strengthen their efforts. That's more
than fine with us. Also keep in mind that as Tyler said, every press
release and FAQ we've produced was reviewed by that group of people. If
someone (who is "Seba"?) decided to forward an Autodesk press release
from Yahoo to /. instead of the open letter, we have no more control
over that any more than we control the MapServer community or the
steering committee. 


MapServer should continue to grow and, believe me, it's in our interest
to see it grow. We obviously hope people will look at the code we're
contributing as well, but we're just happy to be in the open now. 

 

Now that we're past the initial step of disclosure, the discussion
should be as wide open as possible. We are all actively talking to other
entities about sponsoring the MapServer Foundation. Just as we don't
want Autodesk to be the driver of the community we would like to see
more involvement by other players in the GIS world. 

I hope this answers some questions and generates more. I look forward to
your feedback. 

 

Gary Lang

gary.lang at autodesk.com

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/mapserver-users/attachments/20051129/a8b5008d/attachment.html