Re: GPL & commercial software, the critical distinction

[Adam Back <aba@dcs.ex.ac.uk>]

Jim Choate writes:
> It has been asserted that the use of GNU code within a project causes that
> product to be GPL'ed and as a result the GPL is not commercialy viable.
> 
> This is an incorrect interpretation.
> 
> The GPL does say that if you use GPL'ed code in your project then the
> project is GPL'ed, it does NOT say that if you make function calls into a
> GPL'ed library that the product is GPL'ed. 

There are two distinct licenses promoted by the FSF.  They are the GNU
GPL (General Public License) and the GNU LGPL (Library General Public
License).

As you suggest the LGPL is usuable.

However Werner Koch's GNUPG (OpenPGP implementation) uses GPL (not
LGPL) -- or at least it did in version 10-0.0.0, which is the source
tree I have here.

So the comment was on Werner's license, and use of GPL in general for
crypto code, which started this discussion when he offered that it
contained a GPLed Twofish implementation, when Bruce Schneier asked if
anyone had Public Domain implementations in various languages.  (Not
matching Bruce's PD requirement, note).

> This distinction makes fully commercial and source-secure products
> within the GPL infrastructure possible and feasible.

I did make this distinction in my 2nd post on this topic:

: (There is a difference between GNU and GNU Library.  GNU library allows
: you to use a library without infecting your entire software.  GNU
: library is sort of usable.)

Yes.  My point is to highlight for those writing crypto code with the
aim of crypto deployment NOT to use GPL, but preferably (in my view)
PD, BSD, or lastly LGPL.

I went on to suggest half-seriously a "cypherpunk license" which
restricted use of the code to code without government back doors.

Adam