Re: [gentoo-project] [RFC] GLEP 76: Copyright Policy

[Rich Freeman]

On Mon, Jun 11, 2018 at 12:25 PM NP-Hardass <NP-Hardass@g.o> wrote:
>
> On 06/10/2018 04:34 PM, Ulrich Mueller wrote:
>
> > Copyright Attribution
> > ---------------------
> >
> > All files included in Gentoo projects must contain an appropriate
> > copyright notice, as defined by this policy.
> >
> > A proper copyright notice appears near the top of the file, and reads::
> >
> >     Copyright YEARS LARGEST-CONTRIBUTOR [OTHER-CONTRIBUTORS] and others
> >
> > The largest contributor is whatever entity owns copyright to some
> > portion of the largest number of lines in the file.  Additional
> > contributors can be listed, but this is neither required nor
> > recommended.  The "and others" text may be omitted if the explicitly
> > listed contributors hold copyright to the entire file.
>
> Why is this not recommended?

So, I came up with this to try to keep it as simple as possible.

My concern was basically analogous to the BSD advertising clause
problem.  If you start accumulating authors on this line then it gets
unwieldy.  How do you draw the line?

I suggested drawing the line at whoever touched the most lines, mainly
because it is simple.

IMO ANY solution is going to be imperfect, so simple trumps all.

But, it certainly isn't the only possible solution to this problem.

Keep in mind that notice is not the same as
authorship/copyright/credit/etc.  The "and others" is important.
Being #2 doesn't in any way diminish your rights under the law.  The
law simply requires a notice, so we need to come up with one.  It
shouldn't be viewed as "this is the only important contributor to this
file."  If we could not have any notice at all legally then that would
be simpler still.  Git already tracks who did what, and should always
be the place to go for this info.

> If developer A writes 51% of the lines of an ebuild and developer B
> writes 49%, should B not be listed?

Under the policy, no.

> What if all the metadata lines defining variables consists of 75% of the
> file and was written by A, but the core functionality of the ebuild (25%
> by size) was written by B?

Under the policy, "A and others" is listed.

> If A writes an ebuild, and B replaces a majority (>50%) of the ebuild,
> should B remove A from attribution?

Under the policy, yes, assuming this is noticed.  The policy does not
require checking on every commit, but only when the issue is
escalated.  That is actually a change from the wording which tried to
keep things more strict.

> I think that specifying that substantial (though not necessarily
> specific in defining this) contributions/contributors should included in
> the copyright attribution and that substantial contribution attribution
> *is* recommended.

So, the issue then becomes whether we have to define "substantial."
The goal is to have something actionable.  Anybody can run git blame
easily enough.  Figuring out what is "substantial" is harder.

But, the other change to the policy was to relax this and not worry
about keeping it as up to date.  So, in that spirit maybe we can be
more vague and let it be dealt with via escalation.

That seems to be the approach the Linux Foundation takes.  As far as I
can tell they have no policy regarding copyright notice - and the
contents of their files are all over the place.  Presumably committers
make their own judgement calls, and if somebody has a problem with it
they point it out to the Linux Foundation.

That said, the fact that this is basically happened with the eudev
copyright notices didn't prevent it from turning into a bit of a
tempest in a teapot with all kinds of accusations being tossed around.
It was dealt with upon escalation, but people tend to go crazy over
this stuff so some kind of policy that an ordinary dev can understand
wouldn't hurt, so that the issue is prevented.  That was the goal of
the policy.

A big goal here was to keep it simple and understandable but not too
vague.  It shouldn't need constant appeals to Trustees/Council/whoever
to apply to individual situations.

To the extent that the notice doesn't truly give credit to
contributors I'd call it a feature and not a bug.  I'd rather have the
notices viewed as useless for that purpose, because then people won't
constantly squabble about how does/doesn't get included on them.  The
ideal choices would be listing everybody or nobody, but everybody is
cumbersome, and nobody doesn't seem to be legally valid.  So, listing
one by name preserves the nonsensical nature of the notice while still
meeting the legal requirement.  Or something like that...

-- 
Rich