Fredrik Lundh wrote:
Yes, that is what I, and others, have been saying, and doing, all along. Our Mozilla Public Licensed Python application imports (but contains no code from) a GPLed third-party Python module at runtime, but we don't distribute that module, we just tell users to obtain it independently and install it on their systems.Tim Churches wrote:
and how exactly are you going to load a DLL from an EXE file with-
out "mixing, including, or combining" the two?
You can't, but as long as that "mixing, including, or combining" only occurs at runtime, the GPL itself specifically says that is out of scope and the GPL does not apply. Their words, not mine - to quote (yet again):
"Activities other than copying, distribution and modification are not covered by this
License; they are outside its scope. The act of running the Program is not restricted,..."
except that if *you* set things up so the code is combined when run, *you* are copying, distributing, and/or modifying the program in order to mix, include and/or combine your work with the GPL:ed work.
if you leave all that to the user, you're clear.
(relying on word games is a lousy legal strategy in most parts of the world)(Gee, I thought that word games were the entire basis of much legal endeavour in most parts of the world. Patent specifications in particular spring to mind.)
Less flippantly, I agree, and I find it Hanlonesque that the FSF continues to promote a license which is at best unclear and at worst completely contradictory on the issue of whether dynamic, runtime linking or calling does or doesn't fall under its scope. Others have thought the same - as we have seen, Linus Torvalds felt compelled to include an addition to the GPL when licensing the Linux kernel to make it crystal clear that making runtime system calls to the kernel did not fall under the scope of the kernel license.
I would urge anyone who has published a GPLed Python module to add a similar extension in order to clarify their use of the GPL in this regard.
Tim C
-- http://mail.python.org/mailman/listinfo/python-list
