Linking restrictions and shared libraries

Rod Dixon, J.D., LL.M. rod at cyberspaces.org
Tue Mar 13 00:35:16 UTC 2001


You raise a very good question. First, your reference to my Mai v. Peak is
apropos, but section 117 of the Copyright Act is captioned "Limitations on
exclusive rights: Computer programs." Hence, even assuming there is reason
to believe in the continued vitality of the Mai v. Peak holding, section 117
quite specifically narrows the reach of that holding with regard to the
class of cases under section 117(a)(1).

This is not to say that there might not be a set of facts that would fall
outside of section 117(a)(1) involving a GPL'd shared library that permits
dynamic linking to proprietary applications, but those facts would seem to
fall outside of the scope of normal uses of shared libraries, I suspect.

As for derivative works, section 117(a)(1) would be a nullity, if the
copying that occurs by dynamically linking to a shared library (which occurs
as an essential step in the utilization of the computer program)  also
created a derivative work. The access to these works that section 117(a)(1)
provides would be illusory. Hence, the concern here is really with the
narrow class of cases wherein a derivative work *is* created and/or the
library is not intended to be shared or is not part of the OS or API.

Whether the library is intended to be shared should be readily resolved if
its distributed under the GNU GPL or LGPL, so there is no issue there, and
section 117(a)(1) should apply. Whether the dynamic linking somehow creates
a derivative work is a question of law, not contract. In most cases, it is
highly doubtful that the dynamic linking by a proprietary application to a
GPL'd library would result in a derivative work of the Licensor.
Essentially, it is my view that this issue seems to receive far more
attention than is warranted by the number of instances in which a genuine
issue is created. I think it is wise to take the issue seriously, but
reflection should show that the fear of linking is more conspicuous than the
reality of creating undesirable derivative works.

Rod







More information about the License-discuss mailing list