March 22, 1999 -- Richard Stallman has issued an
update to one of his paragraphs and we have appended it below the
article. -- lt eds.
After studying Apple's new source code license, the APSL, I have
concluded that it falls short of being a free software license. It
has three fatal flaws, any of which would be sufficient to make the
software less than free.
* Disrespect for privacy.
The APSL does not allow you to make a modified version
and use it for your own private purposes, without publishing your
* Central control.
Anyone who releases (or even uses, other than for
R&D) a modified version is required to notify one specific
organization, which happens to be Apple.
* Possibility of revocation at any time.
The termination clause says that Apple can revoke this
license, and forbid you to keep using all or some part of the
software, any time someone makes an accusation of patent or
In this way, if Apple declines to fight a questionable patent
(or one whose applicability to the code at hand is questionable),
you will not be able to have your own day in court to fight it,
because you would have to fight Apple's copyright as well.
Such a termination clause is especially bad for users outside
the US, since it makes them indirectly vulnerable to the insane US
patent system and the incompetent US patent office, which
ordinarily could not touch them in their own countries.
Any one of these flaws makes a license unacceptable.
If these three flaws were solved, the APSL would be a free
software license with three major practical problems, reminiscent
of the NPL:
* It is not a true copyleft, because it allows linking with
other files which may be entirely proprietary.
* It is unfair, since it requires you to give Apple rights to
your changes which Apple will not give you for its code.
* It is incompatible with the GNU GPL.
Of course, the major difference between the NPL and the APSL is
that the NPL *is* a free software license. These practical problems
are significant in the case of the NPL because the NPL has no fatal
flaws. Would that the same were true of the APSL.
At a fundamental level, the APSL makes a claim that, if it
became accepted, would stretch copyright powers in a dangerous way:
it claims to be able to set conditions for simply *running* the
software. As I understand it, copyright law in the US does not
permit this, except when encryption or a license manager is used to
enforce the conditions. It would be terribly ironic if a failed
attempt at making a free software license resulted in an extension
of the effective range of copyright power.
Aside from this, we must remember that only part of MacOS is
being released under the APSL--and it is the lowest level part. The
only practical use for this code is to run the non-free part of
MacOS. It will not help free operating systems, because they
already have the low-level drivers for the PowerPC Mac.
Overall, I think that Apple's action is an example of the
effects of the year-old "open source" movement: of its plan to
appeal to business with the purely materialistic goal of faster
development, while putting aside the deeper issues of freedom,
community, cooperation, and what kind of society we want to live
Apple has grasped perfectly the concept with which "open source"
is promoted, which is "show users the source and they will help you
fix bugs". What Apple has not grasped--or has dismissed--is the
spirit of free software, which is that we form a community to
cooperate on the commons of software.
Aside from this, we must remember that only part of
MacOS is being released under the APSL--and it is the lowest level
part. The only practical use for this code is to run the non-free
part of MacOS. It will not help free operating systems, because
they already have the low-level drivers for the PowerPC
Apparently this was not entirely true. People tell me that some
of the information in the released sources has resolved some
remaining uncertainty about the hardware, which has helped the
writing of PowerPC Mac drivers for Linux. It is good of Apple to
make this information available, but this does not affect the main
conclusions of my analysis.
Also, some say that the code released is actually a system which
is sufficient for certain uses (though not including the graphical
convenience that is the Macintosh's main technical feature). If so,
I stand corrected; perhaps the code Apple has released would make
some nontrivial contribution to the free software community, if the
current APSL were replaced with a free software license.