BSD / GPL compatibility - Infection by "Libraries"

Dennis E. Hamilton infonuovo at
Wed Feb 16 22:01:15 UTC 2000

You've asked a key question.

	Welcome to the world of behaving literature

1.1 Based on my readings of discussions on the question of infection by
binding, no one knows for sure.  It is all speculation about libraries,
linking, binding, run-time coordinated use, etc., etc.  There has been no
test in court.

So, what is one to do?

1.2 In the case of GPL (and specifically, what I understand to be the RMS
view of it), one can take a simple principled view that if everything is
GPL'd, then there is never an issue.  That is a perfectly straightforward
approach.  And it is sufficient for the world that RMS is creating to live
in, consistent with what I assess to be a key matter of personal integrity
for RMS.

1.3 There is another view that I favor.  I want to create software that is
widely used and shared.  I don't want users of that software to be concerned
that their dependence on my contribution may place them in a position where
copyright infringement could be claimed because of some weird technicality
around dependence (static or dynamic) of one software element on another.
That is how I have come to favor the MIT-style licenses that go no farther
about derivative works than the OSD requires.  I want people to know they
can safely use the software/code and I want them to feel welcome and valued
in contributing back, but not compelled to do so.  This way, I don't have to
worry about what a library or component (or even service) is and whether
there is an open-source condition on their work as a consequence.


2.1 By not constraining the form or license on derivative works, and placing
simple conditions on provenance and distinguishability (the artistics,
non-confusion stuff), even if a library use or a binding use is found to be
a case of derivation, there is no infringement because adopters already have
a license that permits what they have done.

2.2 If their work is found not to be a derivative, the license is moot.
Either way, the users and adopters are on pretty safe ground and can operate
with some confidence.

2.3 (The SDM license draft I am working on, a "simple" open-work license, is
specific about that whereas the MIT-styles are often more implicit about it
than I'm comfortable with.)

2.4 I looked at the LGPL and realized I simply did not want to go down that
road.  There is way too much technical specificity about how libraries are
tied to software that uses them, and I couldn't figure out a way to abstract
that to some easily specified and readily-applied principles.   For example,
I think using the compiler to do linking is also all right, and I don't want
to go into the hairsplitting that seems to come up here in any case.


For now, I am pretty comfortable with this approach.  I also like the
Open-Source Writers Group approach and how the MIT-like approach fits with
open publication licenses.

When I did that little analysis about the Angels, Borg, Inc., and Cavaliers
last night, I did make myself a little queasy over the appearance of
infringement when extending/maintaining my work makes changes that someone
else has also arrived at in a GPL's derivative.  (I am safer with regard to
the closed-source Borg case so long as I had no way to see their source.)

I am resolved, right now, to simply not fear something that hasn't actually
happened.  I am sure I can deal with if there is ever a concern or dispute
about something like that.  It's not as clear-cut as I would like, but I
still prefer that situation to having people avoid using my work, whether
commercially or otherwise because they are not clear on where they may be in
a grey area over determination of derivative works.  Since I do middleware,
this is not a trivial concern for me.

-- orcmid

Dennis E. Hamilton
mailto:infonuovo at
tel. +1-206-779-9430 (gsm)
fax. +1-425-793-0283

-----Original Message-----
From: David Johnson [mailto:arandir at]
Sent: Tuesday, February 15, 2000 23:36
To: infonuovo at; Dennis E. Hamilton; Ian Grigg;
license-discuss at
Subject: RE: BSD / GPL compatibility - Derived vs. Fair Use

On Tue, 15 Feb 2000, Dennis E. Hamilton wrote:

> 2.	To focus on discussion of derivative works.  Making derivative works is
> right reserved to the original copyright holder, and so a license is
> required to make one.  And this is all provided for under copyright law.
> particular, there is no implied right to make a derivative work by someone
> who is not the copyright holder, so what a copyright license says about it
> is highly pertinent.

In terms of copyright law (and not in terms of programming usage), is an
application that links to a library considered derivative? I would say that
static linking is derivation and runtime linking is not. But I'm not sure
dynamic linking. RMS brings up some good cases against it, but on the other
side there are the arguments that the two are created and distributed
separately, and that the only differences between dynamic and runtime
are purely mechanistic.


More information about the License-discuss mailing list