Subject: Re: For Approval: Open Source Software Alliance License
From: Rick Moen <rick@linuxmafia.com>
Date: Thu, 25 Sep 2003 01:48:43 -0700

Quoting Sean Chittenden (sean@chittenden.org):

> Hrm, that's not the intent, nor how I read it.  "Redistributions of
> source code" means the source code in question that is licensed under
> the OSSAL, not the software that it is linked to.  In the same vein,
> since the LGPL allows closed source applications to be linked with
> LGPL libraries and the LGPL does not require that the closed source
> application have its source published (only the the LGPL'ed library's
> code, which is not the target of the phrase, "source code"), the LGPL
> does not meet this requirement, therefore allowing OSSAL programs to
> link with LGPL libraries.

Your intent might be clearer if you were to add a couple of words 
(shown in emphasis):

     5. Redistributions of source code in any non-textual form
     (i.e.  binary or object form, etc.) may not be linked to
     software that is released with a license that requires
     disclosure of _the OSSAL-covered work's_ source code (ex: the GPL).

An aside:  Linking in object form of (e.g.) an OSSAL-covered work's
source code with GPL-covered source code does not and could not "require
disclosure", for two reasons:  (1) GPLv2 imposes no source-access
obligation based on mere linkage or other creation of derivative works.
That obligation kicks in only upon redistribution of those derivative
works.  (2) More to the immediate point, the GPLed work's author has no
power to compel anything at all concerning access to the other work's
source code.  It's somebody else's property, after all.

A lot of people seem to get easily confused about the latter point,
especially in the BSD camp (**cough** Darren Reed **cough**):  All 
GPLv2 does is state (per section 2 and elsewhere) that, _if_ you cannot
pass along in your redistributed derivative work access rights to the
entire work's source code, then GPLv2 grants you no right to
redistribute the GPL-covered portion at all.  It doesn't change the
licence status of the other portion:  It couldn't; that's simply not
within its powers.

I.e., there is zero risk that GPLv2 could "require" disclosure of 
an OSSAL-covered work's source code:  It has no authority over such
code, so how could it?  (I pointed out to Darren Reed that his
imposition of a similar provision in one recent licence covering his
IP-filtering code was a no-op for this exact reason.)

(Please note that nothing in the above is intended as licence advocacy.
I don't do that sort of silliness.)


> Free software == GPL, right?

Not right.  I'm tempted to do a Venn diagram, but will spare you the
ASCII-graphics torture entailed.  ;->

Free software is a semantic map referring to the same territory as
does open source (after you strip ideological emphasis, etc.).  It can
be broken down into two sub-categories:

Copyleft type: conditioned on source-access obligation.  E.g., MPL, LGPL, GPL.
Non-copyleft type: no such obligations.  E.g., old BSD, new BSD, MIT/X.

> Open Source + product == possible.  Free software + product ==
> non-viable product.

As you'll see from the above explanation, you're committing a category
error.


Returning to the earlier point:

> If the bits are OSSAL, a business can trust on the OSSAL bits always
> being OSSAL.

This would be automatically true by default operation of copyright law,
with or without OSSAL clause 6.  To reiterate:  Licences over other
codebases used in combination with the OSSAL-covered code _could not_
affect licence status of the OSSAL-covered portion of the derivative
codebase.  How could it?  Doing so would violate the property rights of
the OSSAL codebase's copyright owner.

I _think_ I finally got that point across to Darren.  But I see that the
misconception he suffered from lives on.

-- 
Cheers,
Rick Moen                                        This space for rant.
rick@linuxmafia.com
--
license-discuss archive is at http://crynwr.com/cgi-bin/ezmlm-cgi?3