Re: Humor me: Postgresql vs. MySql (esp. licensing) - Mailing list pgsql-general

Randolf Richardson Wrote:

>     Although I view the GPL as well-intended to ensure that free software
> remains that way, I still find that the BSD License seems to be better
suited
> to the needs of businesses at this point in time.
>
As long as we are on the subject of licenses, here is my point of view:
Different licenses for different businesses.  I am also trying to show why
PostgreSQL's licensing puts it in a good position to take advantage of
MySQL's mistakes.

Open source licenses really break down into three main groups each of which
do a good job of serving the needs of certain types of businesses, and each
group has its major success stories...

1)  GPL-- includes a few derivative licenses as well from the FSF and
others.  Best success stories are Linux and the GCC.  I personally doubt
that IBM would be hiring so many developers to work on the Linux kernel of
Sun could take the code and incorporate it into Solaris.  Major business
benefit for the GPL.

2)  BSD/MIT class of licenses.  Success stories include BIND and Apache.
Helps build a wider community of proprietary and open source developers.

3)  Alladin Public License and spinn-offs.  The APL is designed to allow the
software to be used in other open source projects, and be distributed free
of charge.  However, if MAY NOT be distributed for a fee or tied to services
without additional permission from the vendor.  This has the business
benefit of ensuring greater royalites.  There have been several success
stories here (iirc. Ghostscript was once and may still be released under
this license, as is PDFLib).

The big issue with licensing here, however, is the fact that MySQL, by
releasing the client libraries uder the GPL has essentially said that any
developers building proprietary apps must buy expensive licenses from them.
This is similar to what Trolltech has done with QT.  The result is that,
although these products (MySQL and QT) have large open source followings,
the are not making large inroads into corporate space, and will likely never
do so when there are more free alternatives to be had (PostgreSQL, GTK+,
etc).

Now, the MySQL issues can be easily circumvented.  One could relatively
easily write (in PHP), a GPL'd middle component which would provide a simple
SOAP interface for MySQL, and then use proprietary apps to connect to it--
one could even distribute the MySQL client libs without infracting on the
license in that way, but it is too much overhead and quite frankly work when
there are better alternatives.

The GPL was designed with the idea that programs would be relatively
self-contained, and that non-GPL'd programs could easily interact with them.
The other licenses make no such assumption.  And in order to be competitive
in the corporate workspace, GPL'd programs need to be self-contained.

A good example of this problem was a program I have been developign for a
couple years called HERMES.  It is a CRM/ERP platform that (still) supports
both MySQL and PostgreSQL.  The homepage is at
http://hermesweb.sourceforge.net.  The program is licensed under the GPL,
and I do not have the right to change that since others have contributed
code.  The problem is that at the moment, the program is NOT self-contained,
so any extensions, new interfaces, etc. must also be GPL'd.  This severely
limits the community that can find the program useful.  The solution is to
add a set of basic interfaces which will allow non-GPL'd programs to talk to
the server.  The current approach is to create a set of Database Level API's
(Stored Proceedures), LDAP bindings, and web services.  In this way, we hope
to allow the program to become the center of a larger community.

Also, even with BSD licenses, there is a strong incentive to share code,
since it ensures that the burden of maintenance is minimized.  Therefore the
BSD license is not so weak as many GPL zealots would like to think.

The resulting situation is that MySQL has some licensing and technical
issues that make it a very bad fit for enterprise use.  PostgreSQL is both
more free (in that closed source programs can CONNECT to it) and technically
ahead of MySQL.  It is also more rugged and performs better under real
circumstances.  For this reason, I cannot think of a company (aside from web
hosting services) choosing MySQL over PostgreSQL.

Web hosting services are a special exception, and I think that we could
provide better tools for managing hosted accounts.

Best WIshes,
Chris Travers