<?xml version="1.0" encoding="utf-8"?>
html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en-US">
<title>Open Letter to Mr. Larry Ellison Regarding Oracle Corp.'s Treatment of its Free and Open Source Software Assets</title>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<h1>Open Letter to Mr. Larry Ellison Regarding Oracle Corp.'s Treatment of its Free and Open Source Software Assets</h1>
Dear Mr. Ellison,
First of all, I'd like to thank you for the tremendous past and present
contribution that Oracle Corporation has done for the world of Free and Open
Source Software. The Open Source world would be far poorer without such
important open-source contributions such as OpenOffice.org, Java,
Berkeley DB and InnoDB, OpenSolaris, MySQL, many other lesser known projects,
and the many contributions of Oracle (and in the past Sun Microsystems Inc. or
one of the other companies with which Oracle has merged), to global and
successful open-source projects. I'm also in great admiration of your success
as a businessman and entrepreneuer in making Oracle Corp. into what it is
today, which obviously involved a lot of wisdom, capability, courage,
determination, and vision, and which is indicative of a personality
of high merits.
The purpose of this open E-mail message is to make a few concrete suggestions
that will benefit both Oracle Corp., its software and hardware business, and
the global community of users and customers of hardware and software. (As a
practical man, I prefer easy to implement, concrete suggestions, rather than
advocating global "policy" or "attitude" massive changes that can never be
I hope you will consider implementing these suggestions for everybody's
My name is Shlomi Fish, and I'm a longtime user and developer of open-source
software, and an active essayist and writer (or "blogger" possibly) on the
topic. While not a very good businessman myself (and in fact having a poor
track record as far as persisting in workplaces is concerned, where I've only
worked as a "grunt" programmer, so far), I am very interested in the business
and social aspects of open-source software, cultural works, or business in
computing and non-computing topics in general.
You can learn more about me from my home site:
<a href="http://www.shlomifish.org/">http://www.shlomifish.org/</a> .
<h2>What I Don't Suggest Doing</h2>
First of all, I'd like to mention what I don't suggest that Oracle will do,
just to put it out of the way.
<h3>Making the Oracle RDBMS Open-source</h3>
The Oracle SQL relational database and some related products are proprietary,
high cost, products that are very powerful, popular in many contexts, and
are naturally not open-source.
The Oracle SQL server is not the fashion of the month among web-developers,
and other users (who tend to prefer to use such open-source SQL databases as
SQLite, PostgreSQL and MySQL, or even non-SQL databases or a filesystem-based
store), but it is still widely used, deployed, and has many contexts where it
one of the only sane choices.
As a FOSS enthusiast, I'd love to see something similar in scope as
open-source software, but as a businessman looking from Oracle's viewpoint,
I don't see why Oracle should make it open-source, and why it would
make such a big difference for most of Oracle's customers, who are willing to
pay for it. I'm also not on the opinion that non-FOSS software is necessarily
immoral, or that companies should be forced to make their software open-source,
and Oracle has been pretty fair in licensing its database (as all of its
versions are available for a gratis download online, with only some
restrictions for deployment).
So I'm glad we took it out of the way.
<h2>What I suggest doing</h2>
<h3>Make the Solaris Licence GPLv2 and GPLv3 Compatible</h3>
The OpenSolaris licence - the Common Development and Distribution Licence, as
now stands, has been qualifiedas "free software" by the Free
Software Foundation (<a href="http://www.fsf.org/">http://www.fsf.org/</a>),
but is not GPL version 2 and GPL version 3 compatible.
Since GPL version 2 is used by the Linux kernel and by many other projects
and since GPL version 3 is the prefered licence of the GNU project and many
other projects, that means one cannot usually take code from the Linux
kernel and integrate it into OpenSolaris or vice-versa.
If the CDDL licence of OpenSolaris will be changed to a GPL-compatible licence,
then it will allow OpenSolaris to make use of many Linux drivers and a lot
of GNU or GPLv3 software, without facing any restrictions. Being a two-way
street, this would also allow OpenSolaris code to be used in GPLv2 or
GPLv3-licensed projects, but: 1. That is not a bad thing. 2. OpenSolaris will
still benefit a lot from it. 3. It will contribute to Oracle Corp.'s
"karma" among the users.
<h3>Support and promote GNU/Linux and other Open Systems on UltraSPARC servers</h3>
The UNIX world traditionally preferred open-source software (or at
least software with available source) due to the large diversity of its
underlying CPU architectures. For better or for worse, there are a much smaller
amount of underyling architectures in use today, but in the meanwhile, FOSS
turned out to be advantageous in its ability to utilise a very large amount
of contributors (whether paid or working-free-of-charge), who enthusiastically
contribute in many ways. Often, many companies are stakeholders in the same
open-source project, because they all benefit from it.
As a result, free and open-source operating systems such as GNU/Linux have
become much more polished, more powerful and less buggy and quirky than
their proprietary equivalents, and are also what most open-source enthusiasts
are used to and love.
A friend of mine, who had to work with many UNIX and UNIX-like systems
once commented that "HP/UX is not UNIX, and AIX is even less than that.". For
him, the only sane UNIXes were GNU/Linux and Solaris. When I told that to
someone on IRC, he said that he always believed he had originated it.
Like it or not, unless OpenSolaris will really be improved and catches up
with GNU/Linux (see the point above), then Solaris would be soon to follow,
as it cannot catch up with the accelerating speed of Linux's development.
For most purposes, the young people of today want to use tried-and-tested
Linux distributions on their servers: Red Hat Enterprise Linux and its
derivatives, Debian GNU/Linux, Ubuntu - even Gentoo and Archlinux (and naturally
some people prefer the libre BSD operating systems.)
With all the progress in the portability of code written for POSIX-systems on
C and C++ languages, the underlying architecture becomes less and less
relevant, as far as running a server is concerned, and Oracle Corp. and
similar hardware vendors can make a fortune by selling high-end
UltraSPARC-based servers that are designed to run Linux, and where some
versions of Linux is actively supported. With all due respect to the recent
"Webscale", highly distributed, cluster-based solutions, most customers
know well enough that it is a better idea to maintain a small amount of
very powerful computers, instead of 10s, 100s or even 1000s of underpowered
computers, which would be maintenance nightmare and may actually perform
much worse due to the inherent slowness of one's networking.
This is a golden opportunity for UltraSPARC-based machines, which can only be
realised if Oracle actively pledges support for some popular GNU/Linux
-based operating systems.
<h3>TODO: make Java and OpenOffice.org into bazaar-style open-projects</h3>
See: http://lwn.net/Articles/370157/ .
<b>TODO</b> : consider using the X11 licence or a similar licence, so the
Golden Rule of "I can fork and you can fork it too" will be OK. (Copyright
assignment is a problem, but not so much.).
<h3>TODO: avoid going on offensive Software Patents' Actions</h3>