On Mon, Jun 7, 2021 at 11:23 AM Giacomo Tesio <giac...@tesio.it> wrote:

> On June 7, 2021 2:44:56 PM UTC, Jakub Jelinek wrote:
> >
> > Nonsense.  GCC codebase doesn't have a single copyright holder for
> > decades, just look at the source.
> >
> > libffi has various copyright holders
> > include/hsa* has AMD as copyright holder
> > gcc/go/gofrontend and libgo has The Go Authors as copyright holders
> > liboffloadmic has mostly Intel as copyright holder
> > libphobos has mostly Digital Mars (and/or the D Language foundation)
> > as copyright holder
> > libquadmath has Sun Microsystems as copyright holder on various parts
> > libsanitizer has mostly the LLVM authors as copyright holders
> > zlib has various copyright holders
>
>
> The simple fact that you have been able to list the copyright holders
> of the various submodules (as distinct from the rest of GCC under a
> single FSF copyright) shows the advantage of the previous policy.
>
>
> > So, a few extra copyright holders under DCO instead of assignment to
> > FSF will not really change anything significant.
>
> I'm afraid you are being a bit naive here.
>
> You just need one individual who decide to act as "copyright troll" years
> after his contribution has been accepted (things and people change,
> as you know) to cause demage to some users.
>

The copyright troll risk is much, much lower for GCC than for Linux.
First, because GPL3 specifically addresses the over-strict automatic
termination rules in GPL2 that copyright trolls leverage.  And also because
there are many fewer redistributors of GCC, and they are in the business of
distributing software.  If you are redistributing GCC, it's going to be in
some sort of package format that is also a convenient medium for
redistributing the source.  If you aren't redistributing GCC, just using
it, then the GPL places no restrictions on you anyway.

Jason

Reply via email to