The Industry choice

[Paul Rubin]

Jeff Shannon <jeff at ccvcorp.com> writes:
> Note that the so-called 'viral' nature of GPL code only applies to
> *modifications you make* to the GPL software.  

Well, only under an unusually broad notion of "modification".  The GPL
applies to any program incorporating GPL'd components, e.g. if I
distribute a Python compiler that incorporates some component from
GCC, then my entire Python compiler must be GPL'd even though the GCC
component is isolated inside the Python compiler and I wrote the rest
of the Python compiler myself.  If I don't like this, I have an
obvious recourse, which is don't use GCC components in my Python
compiler.

The notion here is that the GCC components are attractive enough that
being able to use them provides an incentive to GPL my Python
compiler, which I might not do otherwise.  

> (Problems may come if someone licenses a library under the GPL; that's
> what the LGPL was invented for.  But the issue here is not that the
> GPL is bad, it's that the author used the wrong form of it.)

The "problem" is not a problem except that in the case of some
libraries, simply being able to use a library module is often not
enough incentive to GPL a large application if the library module's
functionality is available some other way (including by
reimplemntation).  If the library does something really unique and
difficult, there's more reason to GPL it instead of LGPL'ing it.

> The 'infective' nature of the GPL *only* comes when you make use of
> the *extra* privelidges that open source grants.  So yes, those extra
> privelidges come with a price (which is that you share what you've
> done); but if you don't want to pay that price, you have the choice of
> not using those privelidges.  This does not, in any way, prevent you
> from using GPL'ed software as a user.

Well put.