Subj : Re: GNU Public Licences Revisited (again)
To   : comp.programming
From : Arthur J. O'Dwyer
Date : Tue Aug 23 2005 10:47 pm


On Tue, 23 Aug 2005 gswork@mailcity.com wrote:
> Arthur J. O'Dwyer wrote:
>>    Writing a bunch of extra code that calls the library /is/ altering the
>> library, from the literal-minded CS point of view. You're taking one
>> collection of source code and turning it into another, different
>> collection that includes a lot of the original. That makes it a derivative
>> work, copyrightically speaking. If the library was GPL'd, the resulting
>> work must be GPL'd. If the library was LGPL'd, the resulting work doesn't
>> need to be GPL'd, although it can be, of course.
>
> that's interesting - is it a sliding scale of 'involvement' or an
> absolute,

   It's all based on U.S. (and perhaps international) copyright law, which 
as I'm sure you're aware is an utter mess.  So the "involvement" is one or 
more of:
   (1) reproduction of the original
   (2) creation of derivative works based on the original
   (3) distribution of the original
   (4) public performance or display of the original
As you can see, these are fairly Stone-Age concepts that apply only 
metaphorically to software. For example, I think it's been established 
that you can legally read a copyrighted e-book, even though a 
"reproduction" of the text has been made in your computer's RAM, and 
another in video memory, and another, fleetingly, on the screen.
   In your case, we're talking about the creation of a "derivative work."

> [Suppose we only call the library, and don't modify it.]
> that being the code which calls the library.  It does I suppose use the
> code and link it into binaries - but is that all it takes to oblige the
> entire project to bcome open source or am i just missing a point here?

   If you take the code and link it into your binary, that's definitely a 
"derivative work" from the standpoint of the GPL, and you'd have to GPL 
it. If the library were LGPL'd, I think you would *not* have to. (LGPL is 
a "weaker" version of the GPL, meant mostly for code libraries. I don't 
know much about it.)
   If you dynamically load it at runtime, like a DLL or something, then I 
believe the answer is hazier.  Ask a real expert, or an IP lawyer.


> I'm not sure that GPL would automatically lead to ever more open source
> by 'forcing' other projects to become so, or whether infact it would
> just lead to seperateness - GPL gets used in GPL and other things get
> used in closed source.

   Yes, that's actually what's happening in practice all over the place. I 
know from experience that Microsoft, for example, is so GNU-phobic they 
won't even let programmers examine any sort of open-source code, GPL'd or 
not, lest it infect their IP.
   Stallman would probably say, "Good, that's going to hurt them."  In 
practice, it really doesn't seem to; in fact, the virality of the GPL 
probably hurts most for the independent developers who want to be 
open-source for security and hacker-friendliness, but /not/ GPL.  Which 
again would make Stallman say "Good, that'll make them want to be GPL." :)


> Are there any test cases setting the precedent?   It would be
> interesting to see the GPL in legal action.

   I seem to recall there was one once, but I don't know.  That really is a 
question for gnu.whatever the last guy said --- I'm sure someone who reads 
that newsgroup must study that sort of thing.

HTH,
-Arthur

.