[License-discuss] GPL and proprietary WebAPIs

[Clark C. Evans]

On Sun, Dec 25, 2011, at 01:01 PM, Rick Moen wrote:
> Quoting Clark C. Evans (cce at clarkevans.com):
> 
> > What confuses me and what I'm asking here is why licensing 
> > professionals focus on two items that I consider irrelevant: 
> > (a) what is the type of linking between D and P?
... (b) is D a derived work of P?   What we know is that D 
... is derived from O and that D requires P for its operation.
>
> Please name one such licensing professional (who is not an FSF
> spokesman).

What was on my mind was this journal article:

 Software Interactions and the GNU General Public License
 By Malcolm Bain in IFOSS L. Rev. Vol 2, No 2 (2010) 
 (http://www.ifosslr.org/ifosslr/article/view/44)

What was also on my mind was an informal side chat
with an attorney on the stack overflow question [1].
I was referred to the GNU FAQ, especially the answer
for plug-ins [2] where the applicability of the copyleft
"depends on how the program invokes its plug-ins" and
aggregation [3] where it says "sockets...  are normally
used between two separate programs".  This attorney
said the specific details of the case were necessary
to give any advice, but after my insistence, commented
that the community consensus is likely correct.  

Probably you consider both of these professionals to 
be associated with the free software foundation.  For 
what it's worth, our own corporate attorney doesn't 
have a concern with regard to linking technologies, 
he advices that GPL is quite vague and if we were to 
use it, we should publicly exclaim our interpretation.

To this regard, Larry Rosen's comment [4] is refreshing,
matches my expectations, and is broadly helpful: 
 
 "If GPL advocates insist upon distinguishing 
  among types of functional linking, then talented 
  software engineers will avoid disputes by 
  building shims, APIs, or use dynamic linking to 
  accomplish their functional goals."

Unfortunately, this doesn't give me enough guidance on
the applicability of copyleft; specifically with my
3-work scenario where someone uses a shim/adapter to
include proprietary functionality via a WebAPI.  I keep
asking this in this public forum since it matters quite
a bit to the effectiveness of the GPL.  If this sort of
work around is acceptable by the community, then other
than nuance value, there really is no point in copyleft, 
and one should use Apache or keep the work proprietary.

Assuming a shim/adapter creates a derived work under 
copyright law so that its distribution could be limited
based on conditions of the license agreement, does the
GPLv3 actually prevent the distribution of a derived work 
when its operation requires an independent, proprietary 
component that has no free software substitute?  

In particular, does the relation vis-a-via copyright law 
between the adapter/shim to the independent proprietary 
work matter if it's clear enough that the adapter/shim 
cannot operate without the proprietary work? 

Best,

Clark

[1]
http://stackoverflow.com/questions/4351119/if-i-use-gnu-gpl-code-with-my-own-server-side-code-do-i-need-to-open-my-server/
[2] http://www.gnu.org/licenses/gpl-faq.html#GPLAndPlugins
[3] http://www.gnu.org/licenses/gpl-faq.html#MereAggregation
[4]
http://projects.opensource.org/pipermail/license-discuss/2011-December/000091.html