On Dec 7, 2003, at 17:07, Arnoud Engelfriet wrote:
Anthony DeRobertis wrote:
Huh? Please, could someone please find the derivative works in the
following, in chronological order:
1. I create a program, Anthony's Foo Editor, and add a plugin
API.
I release my program under the MIT X11 license.
2. Weston Manning (a new maintiner) uploads Anthony's Foo Editor
as
afe.
3. Marc Spencer creates a plugin, Frobit, under the OpenSSL
license
4. Weston Manning uploads afe-frobit
5. Duncan Finch creates a plugin, Barnitz, under the GPL,
version 2
6. Weston Manning uploads afe-barnitz
If I understand the FSF correctly, they claim that a package
containing both 'afe' and the 'barnitz' plugin is a derivative
work of the 'barnitz' plugin.
No package containing both was created in the above!
Even if one were, it'd be a compilation --- not a derivative work ---
as there was no modification of either work. IOW, a mere aggregation.
And since the FSF's logic is "linking at runtime means
derivative work before runtime", it follows that the bundle
is a derivative work of the plugin.
That doesn't follow. If we assume linking at runtime means creating a
derivative work before runtime, then we can conclude only that the
plugin is a derivative work of the plugin host.
So, I must distribute the complete source code of the plugin available
under terms compatible w/ the GPL. The X11 license is one such set of
terms.
If I make a derivative work, app + plugin A, and a work, app + plugin
B, then that doesn't show that A is a derivative of B or vice versa.
So A's license doesn't matter to B and vice versa, unless there is some
other reason for them to be derivatives of each other.
