This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: SGI releases IA64 C C++ and F90 compiler under GPL
In article <3927B5E3.445E9B5@moene.indiv.nluug.nl> you write:
>Linus Torvalds wrote:
>
>> In article <20000516031448.H13075@Jeffreys.suse.de> Phillip Thomas writes:
>
>> >So the real question seems to be, who's going to convince SGI to do the
>> >assignment, ain't it?
>
>> In short, the assignment itself is of quite questionable morality in the
>> first place, and then the FSF/Cygnus has the _gall_ to lay the blame for
>> the trouble their own inflexibility causes on somebody else.
>
>Ho, ho - lets try to stick to the *question* we were trying to resolve,
>namely whether it would be possible to use code from a GPL'd work that
>was *not* assigned to the FSF in one that *is*.
As far as I know, the reason asserted is that FSF lawyers say this is
needed to simplify legal battle.
There is a very unfortunate side-effect though: in usual life,
copyright is distinct from licensing rights.
For instance, Perl is distributed under a dual License: either GPL or
Artistic.
Giving away your copyright to the FSF means the code will be distributed
under the GPL only.
I can see why SGI would like not to lose the ownership of the code.
If they want to dual-license it at some point, say keep a GPL license,
and reissue it under a BSD license, they can---as long as they haven't
forfeited the right to.
Personally, I would prefer a simpler world, where the FSF would not demand
copyright ownership. I have some code (leaner, meaner faster hash-table)
which could benefit gcc, which I am quite willing to release under the GPL,
but which is also under a BSD license. In short, I can't give it to the FSF.
Being your typical hacker, I'm loathe to rewrite that code, just so that it
can have two distinct free licenses.
I'm reasonably certain I'm not alone. This problem might even explain why
some Net/Free/Open BSD code never finds its way into the FSF repository,
even though we don't really want to maintain the code on our own.
(of course, that hash table thingy is anecdotic. I'm certain there is more
heavy-weight, independent code Out There that could be integrated into gcc,
and benefit it, if it weren't for that `GPL or die' aspect of gcc development)