On 12/05/2012 09:24 PM, Richard Fontana wrote:
On 12/04/2012 11:29 PM, Theodore Ts'o wrote:
> On Tue, Dec 04, 2012 at 11:05:57PM -0500, Richard Fontana wrote:
>> (iv) the Source Code of shared
>> libraries that the Covered Work is specifically designed to require,
>> such as by intimate data communication or control flow between those
>> libraries and parts of the Covered Work;
> Are you going to define what is meant by "intimate data communication
> or control flow"? I could imagine that being considered somewhat
> vague.... and things which are vague that can only be settled by a
> lawsuit is, in my mind, unfortunate....
I've now deleted this.
Good-Faith Partial Harvey Birdman Rule Violation Cure:
I recently had a telephone conversation with bkuhn in which we briefly
discussed some of the substantive issues raised in this thread or
subthread. I had thought that bkuhn was going to cure this Harvey
Birdman Rule violation, and I believe he did take some notes for that
purpose, but he hasn't yet posted anything on this list or
communicated otherwise. My recollection of this conversation (which
took place last week) is beginning to fade.
One of the issues we discussed (though only for a minute or two) was
my decision to delete this portion of the Corresponding Source
definition that Ted Ts'o had highlighted. The substantive point I wish
to HBR-ify is that bkuhn observed that an arguably nice aspect of the
deleted clause is that it codifies traditional FSF interpretation. I
believe I pointed out that while I agree that this is a nice feature,
I am not sure that the language of the clause matches such
interpretation as closely as it was intended to do. I pointed out to
bkuhn that perhaps something like this language could instead go into
the definition of 'Derived Work'.
I noted to bkuhn (though not in the following level of detail) my
recollection that during the drafting of GPLv3, in one or more
meetings of 'Committee B' there were some persons who proposed to FSF
counsel that what they saw as insufficient clarity around the scope of
GPLv2 'derivative work'/'work based on the Program' might be resolved
by moving elements of the quite detailed Corresponding Source
definition into a definition of either the successor label to
'derivative work' or else the definition of the basic copyleft
condition in GPLv3 section 5.
The perception was that the definition of 'Corresponding Source' was a
sort of misplaced attempt to define what GPLv2 means by 'derivative
work'. I actually don't think that is quite correct, but it is
important to know that the argument was made and I do believe it
contains an important insight.
What is interesting about the GPLv3 ancestor of the deleted clause is
that it appears to avoid the traditional tiresome legal debate about
GPLv2, shared libraries and 'derivative works' by instead focusing on
what one has to provide as a matter of source code if one distributes
(conveys) a binary.
However, I am not convinced that the result of applying the deleted
clause adequately captures either pre-GPLv3 FSF interpretive tradition
or intuition among most strong-copyleft-supporting, GPL-using
developer-licensors who have been strongly influenced by FSF
interpretive tradition. I do not have time to go into this, but I will
note that the suggestion made some time ago by a person on github (see
thread "Allow linking with any FSF/OSI approved license" from
September) may be worth thinking about here. Ignoring the details of
the suggestion, I believe the suggestion may express an intuition
common among GPL licensors: what matters is not pure orthodox GPL
FLOSS license compatibility (under traditional FSF-ish strong copyleft
interpretive doctrine, however one understands that) but rather
avoidance of certain cases of GPL/proprietary code coexistence assumed
to be allowed under weak copyleft licensing regimes.