Who can help with the GPL v3 effort?
Stephen J. Turnbull
stephen at xemacs.org
Tue Mar 2 23:38:27 EST 2010
Ben Wing writes:
> On Tue, Mar 2, 2010 at 1:00 PM, Jerry James <james at xemacs.org> wrote:
> > The goal is in sight and ... and .. and I can't quite get us
> > there.
I think we *are* there, between the work you've done and Ben's
analysis. Thank *you* very much for the warning about your (lack of)
> > Problem files:
> > 1. etc/editclient.sh
This is small and obvious enough that we can either delete it (my
preference, as long as we solicit contributions to make the xemacs
binary DTRT) or keep it.
> > 2. etc/emacskeys.sco: I think we should delete this.
> > 3. etc/emacsstrs.sco: Ditto.
> > 4. man/cl.texi: License requires us to have the same GPL section as the
> > original, which didn't have a GPL section (as near as I can tell).
Move it into lispref, which does. That code is a first-class citizen
of XEmacs and deserves first-class documentation. Unless Mike
objects, but he has been at least moderately supportive of other
initiatives to bring XEmacs Lisp closer to Common Lisp.
> > 5. netinstall/reginfo.h: Waiting to hear from Volker Zell or someone else at
> > Cygwin about whether we can just delete the entire netinstall directory.
> > 6. netinstall/version.pl: ditto
Let's delete it, after tagging a recent version to make diffing easy.
> > 7. tests/Dnd/droptest.el
> > 8. tests/Dnd/droptest.sh
Let's delete them, ditto tagging.
> > Problem files that have equivalents in GPL v3 Emacs:
> > 1. lib-src/digest-doc.c
> > 2. lib-src/emacs.csh: Emacs version is etc/emacs.csh
> > 3. lib-src/hexl.c
> > 4. lib-src/sorted-doc.c
> > 5. lib-src/vcdiff
Non-problems because of the GNU versions AFAICS.
> > 6. lwlib/lwlib.h
> > 7. lwlib/lwlib-Xm.h
> > 8. lwlib/xlwmenu.h
> > 9. lwlib/xlwmenuP.h
I'm happy with the claim that they are covered by the Lucid GPLv1 or
later license or contributions by authors who have given us blanket
permission to relicense their contributions as necessary to allow GNU
> > 10. man/texinfo.tex
> > 11. man/texinfo.texi
I forget the problem with these. If it was a real problem, we can
delete references to part of XEmacs (better, deny it) and claim mere
aggregation in the distribution.
> > 12. man/widget.texi
Delete or copy GNU's.
> > 13. src/s/aix4-2.h
> > 14. src/s/freebsd.h
> > 15. src/s/hpux11.h
> > 16. src/s/irix6-0.h: Emacs version is irix6-5.sh
> > 17. src/s/netbsd.h
> > 18. src/s/openbsd.h
> > 19. src/s/sol2.h: Emacs has sol2-3.h, sol2-4.h, sol2-5.h,
> > sol2-6.h, and sol2-10.h, some of which are explicitly GPL v3
> > or later, and some of which have no explicit license
> > statement.
> > 20. src/s/usg5-4-2.h
I'll defer to Ben on these, he seems to have a pretty complete
> For many files, there was nothing left; for a few others, there was
> only one line or so. The only real problem is sol2.h -- our version
> is totally different from GNU, and there is a fair amount of stuff in
> it that isn't in GNU. Some of that I could trace to Martin, which
> makes it OK, but there's a big chunk of prototypes
Prototypes as such are not a copyright problem. Consider removing or
changing everything that can be changed (ie, comments and argument
names). What is left (argument types and external identifiers) is not
"expressive content" because there's only one right way to do it.
Therefore it is not copyrightable in the first place. Does that get
us to zero? If so, I can't imagine anybody (or court) bitching if we
leave them as they are, because it's trivial to fix.
> that has no obvious contributor and is apparently required for
> Solaris 6, but stopped being necessary some time prior to Solaris 10.
If there's still stuff needed, can't we live with partial support for
Solaris < v10 for a while? In the interim we can get the needed info
from testers and other sources and rewrite anything we delete.
> Stephen, since you are the one most likely to have a fit if I do
> something that you don't consider kosher, what do you think about
> this plan?
I'm fine with it, as you see above.
> Also, isn't there a rule at FSF that contributions less than 10 lines
> don't require copyright assertion?
The rule of thumb for trivial changes is 16 lines, IIRC, but the
problem is that this is *per author*, not per change. Since we don't
have authors, we have to get the *total* suspicious lines down to 16
to invoke this.
More information about the XEmacs-Beta