svn commit: r356142 - in head/sys: dev/ofw sys

[Warner Losh]

On Fri, Dec 27, 2019 at 10:55 PM Rodney W. Grimes <freebsd at gndrsh.dnsmgr.net>
wrote:

> > On Fri, Dec 27, 2019 at 10:27 PM Rodney W. Grimes <
> freebsd at gndrsh.dnsmgr.net>
> > wrote:
> >
> > >
> > > > > You can choose your own license for original work, sure, but
> > > obliterating
> > > > > parts of an existing license by applying a second license which is
> in
> > > > > conflict is probably a poor idea.
> > > >
> > > >
> > > > We don't do that at all: pretty clearly there is no conflict between
> > > > both licenses as you can comply with both.
> > >
> > > The only way to comply with both is to comply with the full 4
> > > clause license.  Hense the 2 clause is pointless in being there
> > > and can never apply until all 4 clause authors agree to change
> > > to 2 clause.
> > >
> >
> >
> > Until such time as Jeff finishes rewriting the files, then we just nerf
> the
> > 4 clause one as no longer relevant since it describes no code in the file
> > anymore...
>
> Slippery slope as that would require a very detailed audit to
> make sure at no time in any way did Jeff or anyone else copy
> or retain any original code.


One we've  done dozens of times in the project's history. People rewrite
things all the time. From the tty layer to things in the vm.


> >
> > We've done exactly the thing Jeff did hundreds if not thousands of times
> > already in the project in code spanning
> > at least the last 25 or so years...
>
> I have to call BS on that claim, the project is just barely past
>

Please watch the tone of your replies. This is not an acceptable tone.


> 25 years old, and we certainly did not do any of this at that
> time, and further the 3 clause came into existance in 1999, and
> the 2 clause was that same time frame, so possibly 20 years.
>

We have several files with 2 clause that date to 1996 (look at many of the
elf_machdep.c files have this date). Both FreeBSD and NetBSD used 2 or 3
clause licenses well in advance of the regent's letter...


> Please show me the 100 to 1000's of files that this occured in.
>

sys/kern alone has many of them, though this sort of thing is hard to grep
for. sys/arm/arm has some. sys/mips/mips has some more. Many with dates
going back at least 15 years. sys/arm/arm/support.S has one that has 3
different sets of clauses, the most recent of which is 2004, the earliest
1997 ( NetBSD, Wasabi and Olivier Houchard).

A grep of the kernel shows ~200 .c, ~20 .s and ~80 .h files that have
multiple licenses, though grep is the wrong tool to know how many are
identical and how many vary. A quick audit  suggests maybe 5-10% of
these are likely to vary.  so not hundreds or thousands, but not zero
either.

I've not looked at userland at all with this quick grep.

>  Not sure why it's coming up now over an annotation that has a
> > specific meaning that's clear and well defined.
>
> No one pendantically legal has been watching commits for 20 years
> is probably why?
>

What about the many legal reviews done by companies that produce FreeBSD
products over the years. Those generally flag things like the beerware
license, but not this detail....

Warner