Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!husc6!contact!umb!karl
From: karl@umb.umb.edu (Karl Berry.)
Newsgroups: comp.lang.misc
Subject: The GNU license.
Message-ID: <873@umb.umb.edu>
Date: 21 Jul 89 21:15:53 GMT
Organization: UMASS-Boston, Boston, MA
Lines: 47


Peter da Silva asks about the licenses for Bison and GCC being
different, and brings up a point about how small the GNU-made code
sequence must be for the work to qualify as ``derived''.

I believe that the following is accurate, but it may not be. I am
neither rms nor the FSF's lawyer.

David Keppel has it right about Bison; because the result contains a
substantial amount of GNU code that did not come from the original
source, the result is copylefted, and must be freely distributable.
Peter's remark that ``this makes Bison useless for commercial software
developers'' is true (at least as long as commercial software developers
don't want to let people see their source). But rms started GNU to
promote free software, not to help people who want to hoard software.

GCC, on the other hand, produces almost no code that is not derived from
the original source. The exception is gnulib. Its routine are
explicitly not copylefted, because they are too small to possibly
qualify for copyright. When the GNU runtime libraries are released, a
final binary that incorporates the libraries will have to be freely
distributable, just like Bison. (Yes, a hoarder can get around this by
distributing only object code for his/her program, source for the GNU
libraries, and having the user do the link. Let's not get sidetracked.)
In other words, the result of a compilation with gcc -c is not
copylefted.

(Emacs, just to forestall any queries about this (this whole discussion
has just taken place on gnu.gcc; perhaps you can retrieve the archives
from prep.ai.mit.edu), produces nothing that is not derived from the
original source. That is, editing a file with Emacs doesn't somehow
make the file FSF property. That would be absurd.)

Finally, Dave Sill says that ``Stallman and the FSF have better things
to do than sue people who try to use their products.''  If someone bases
a program on GNU software, and does not make the result freely available
under the copyleft, but rather hoards it then it is likely that they
will be sued. rms has said this many times. (I believe that NeXT has
developed a C++ compiler using GNU software, and is now trying to get
out of the copyleft, so this is not entirely an academic point. I hope
that NeXT decides to make their compiler freely available, if they
haven't already.)

I hope this clarifies the situation. Remember, this is not a definitive
legal document :-).

karl@umb.edu    ...!harvard!umb!karl