Re: IBM doesn't like the GPL

[Hyman Rosen]

Rahul Dhesi wrote:
> I did not find the phrase "derivative work" in the GPL v3 text.  I found
> it in v2, but you are discussing v3, are you not? If so, it would be
> better to stick to the language in GPL v3.

I am discussing the GPL in general. The power the GPL gets over
a work as a whole comes from copyright law. Therefore, we must
discuss when the GPL can acquire such power.

When a program is linked dynamically against a GPLed library, the
program does not contain a copy of the library. Thus, the only way
that the program might fall under the power of the GPL is if the
program could be considered a derivative work of the library. That
term, "derivative work", comes from copyright law.

But a work is a derivative work of another only if it is transformed
version of that work representing a significant work of authorship,
which is manifestly not the case for a program using a library.

Therefore the license of the library is immaterial to how the program
which uses it may be distributed, and so a program may dynamically link
to a GPLed library without that alone causing the program to fall under
the GPL.

For Java, where there is no static linking, that means GPLed class files
may be used in non-Free Software "programs" (that in quotes because a
Java program is an assemblage of class files, not a unified whole).