RE: Clear up my License information

> Patrik Stridvall <[email protected]> wrote:
> > > Well, we are both right to a certain extent :)
> > >
> > > The problem is that if your patch contains a modified line
> > > that is only in the LGPL WineHQ tree, the 'viral' nature 
> of the LGPL means
> > > that one LGPL line causes the LGPL to apply to the whole 
> patch, and thus
> > > cannot be duel-licensed (without permission of the author 
> of the oiginal
> > > LGPL'ed line). How many times can I use the word LGPL in 
> one sentance? Wow.
> > > :)
> > 
> > Note quite. First of all the LGPL is not quite that viral 
> as the GPL.
> > 
> > The LGPL (unlike the GPL) even with its most restrictive 
> interpretation
> > only applies to code that is either directly derived from 
> LGPL code or
> > code that the LGPL work depends on for it functionallity.
> 
> This is patently false.  If this were true, one could 
> statically link against
> a LGPL library and not be affected.  

Oh, sure you have to be able to relink the work as well,
but that is not really relevant for patches since they usual
works on the source code level of things, so I didn't meantion it.

> Besides, it is the 
> meaning of 'derived'
> which causes most of the consternation anyway.

True. But for patches that "almost" apply it should be
a problem. See below.
 
> > So no, save the work nessessary for Rewind to make any 
> patch from X11
> > licenser apply I don't see any problem from LGPL 
> "contamintion" of the patches.
> 
> This arguable.  Take the following  case:
> 
> Original X11 code:
> 
> foo()
> {
>   code();
>   code1();
> }
> 
> 
> LGPL mod:
> 
> foo()
> {
>   code();
>   if( cond )
>      code1();
> }

Too trivial modification to be covered by copyright at all.
But it doesn't really matter since the idea or fact that an
if is needed is unprotected.
 
> X11 contribution:
> 
> foo()
> {
>   code();
>   if( cond ) {
>     x11_added_code();
>     code1();
>   }
> }

The patch would mismatch here so human intervention would be needed.
 
> The funny thing is that it can be argued that if you don't 
> use K&R styling,
> you'd probably be safe, as the patches would not have any reference to
> the LGPLed "if".  In this case, the "if" could be considered 
> to contaminate
> the patch, 

No it wouldn't. The patch wouldn't apply directly so any added if to the
Rewind
tree would be because of the mergers "unique" expression using the
unprotected
fact or idea that an if is needed. Sure the function in LGPL tree and the
X11 tree
would likely be the same after the merge but that is not a problem. The
similarity
is only because the problem was solved using a standard programming method
based
on the same fact and ideas.

In short:
Only "unique" expression is protected not facts, ideas or methods.

> in the sense that you cannot say the patch was 
> X11. All that
> you can (and should) say is that your contributions are licensed X11.

As I sidenote I don't usually consider the context of the patch as truly
belonging to the patch. 

But there are slighty problematic special cases like whitespace only
modification so if you want to be crystal clear what on what you mean
then yes.

However I think people applying patch to Rewind is clever enough to
realize this regardless, so I see no need to bother.

> > In any case, as I said earlier this is Rewind problem, not 
> the X11 licenser.
> 
> Depends on your perspective.  If your attitude is "Rewind can 
> make use of this
> if it can", then yes, your statement is true.  If you want to 
> make a direct
> contribution to the X11 tree, then the patches should 
> probably be off the
> X11 base.

I think most X11 licensed patches would be useful to them, they
just have to be a little careful but the possible pitfalls is AFAICS
few and obvious.

But for parts where the trees have diverged a lot it would presumable
take some time to integrate the patch eventhough I guess it would almost
always be quicker than rewriting from scratch.