> Btw, we do seem to have code in GCC that is not copyrighted by the FSF. > For example I don't think the FSF owns copyright on the ACATS > testsuite, and libffi mentions (c) by RedHat. For GCC as a project it > should matter that the code is distributable under GPLv3 which I think > Apples changes are.
I thought the point is that Apple WON'T go to GPLv3. But I also think there are other cases where some code isn't under GPL and/or not assigned to the FSF. Things like crt0.c are derived from BSD code and are under the modified BSD license and there's also some public-domain code involved. > Copyright assignment to the FSF might be > convenient, but isn't legally necessary (only necessary by policy and > even that seems to have existing exceptions). I think there's a difference between the compiler proper and libraries and test suites. Because the latter are often derived from other code, my understanding is that the policy on license and copyright holder is less strict than the compiler itself, which doesn't have that problem.
