Subject: Re: For Approval: Microsoft Permissive License
From: Chris Travers <chris@metatrontech.com>
Date: Sun, 19 Aug 2007 15:43:41 -0700
Sun, 19 Aug 2007 15:43:41 -0700
First the general points, then how they affect the MS-PL.

IANAL, TINLA, YMMV, etc.

Donoan Hawkins wrote:
>
> Excluding linking for a moment:
>
> If you only convey your derived work in source code form, with an 
> optional dependency available via conditional compilation, I would 
> agree with you. Your work compiles (without the optional component), 
> but has bindings to allow someone to easily enable support for the 
> optional component. I don't see a problem there.
>
> If you convey in executable form, then your "optional component" isn't 
> optional anymore. You either compiled it in or you didn't, and the 
> answer to that question determines whether you have to provide the 
> component source under GPL or not.
So consider the following (why this is messy):

Libpq compiles optionally with OpenSSL (old-BSD licensed) and is under a 
new-BSD license.  It connects over a network socket to PostgreSQL.  When 
OpenSSL is available, SSL-encrypted network connections to the db server 
are possible for any connecting application.  Even if libpq is compiled 
with OpenSSL support, unencrypted connections can still be used by the 
library if they are not requested by the server.  Hence the absence of 
the library does not prevent most other applications from functioning 
under the most common configurations.

FreeRadius is under the GPL.  Suppose they ugprade to GPL v3.

Suppose there is an LGPL plugin to FreeRadius which uses libpq to 
authenticate users against PostgreSQL.

Now, by your logic, if I am an Linux distributor, I cannot distribute 
Libqp compiled with OpenSSL at the same time that I distribute the 
FreeRadius plugin for PostgreSQL because the corresponding source would 
include OpenSSL and the licenses are not compatible.  This is not 
hypothetical.  Istr Debian refusing to distribute a similar plugin under 
the same concern.

My analysis would be however that the Freeradius plugin under GPL v3 
might be different because of the fact that the plugin does not fall 
under the category of "dynamically linked subprograms that the work is 
specifically designed to require."  This would, in my mind, let a 
distributor off the hook because it draws a line based on a "designed to 
require" standard.

My reading of this is "if the program runs without it, you don't have to 
provide the source under the terms of this license."
>
> Linking is another matter, and I'll defer to others on that since I 
> think some people disagree with the FSF on that point.
>
Linking is not mentioned in the GPL v2 at all last time I checked.

Dynamic linking is specifically mentioned in the GPL v3 as a part of the 
corresponding source, but only in cases of "dynamically linked 
subprograms that the work is specifically designed to require."  I 
believe static linking is implied in other portions of the definition of 
corresponding source.
>
>> Also "corresponding source" includes a number of loopholes which 
>> could be used to exclude arbitrary components.  For example, I 
>> *could* make a Linux device which used closed source libraries as 
>> part of an authentication system, create a GPL'd application for that 
>> device which used those same libraries, and exclude them as "system 
>> libraries" (because they interface with the authentication system 
>> which is arguably a "Major Component").  If you want to use the 
>> application on a different Linux device, you have to pay for the 
>> required libraries.
>
> GPL v3 says:
>
> "A "Major Component", in this context, means a major essential 
> component (kernel, window system, and so on) of the specific operating 
> system (if any) on which the executable work runs, or a compiler used 
> to produce the work, or an object code interpreter used to run it."
>
> So your authentication system is not a "Major Component"...it is not a 
> component of the OS, nor is it a compiler or object code interpreter.

So PAM and NSSWITCH are not a part of the OS?  That is news to me.

I would think that the authentication subsystem would be as much a major 
component as, say, X.org.
>
> All this exception is saying is that you don't have to give people a 
> copy of the source code to gcc and linux just because you use printf() 
> and fork(). Or worse, give everyone a GPL-licensed copy of the source 
> code to MS Visual C++ and MS Windows because you use CString and 
> CreateFile().
>
But how do you define the OS?  Certainly the device in question might 
not include any GNU tools and hence not even be something RMS would call 
"GNU/Linux" but does that make it less of a system component?

My definition of "Major Component" would be a component similar to the 
kernel or windowing system which provides basic services to a 
substantial class of application on the system.  THis seems to me to 
most closely approximate the wording of the GPL v3 since it doesn't seem 
to define this very well.

I see no reason that X.org would be a major component and Oracle, or in 
this case, some hypothetical PAM replacement, would not.
>
>> Does the MS-PL place any specific requirements on the code that the 
>> GPL does not allow?  If so, what are they exactly?
>
>> From MS-PL:
>
> "If you distribute any portion of the software in source code form, 
> you may do so only under this license by including a complete copy of 
> this license with your distribution."
>
>> From GPL v3 section 5c:
>
> "You must license the entire work, as a whole, under this License to 
> anyone who comes into possession of a copy.  This License will 
> therefore apply, along with any applicable section 7 additional terms, 
> to the whole of the work, and all its parts, regardless of how they 
> are packaged."

So, are there cases where you cannot do both?

It seems to come back to the same question we have been arguing on 
another thread.

Suppose I take BSDL code and include it as whole files into my GPL 
application.  Do those files as distributed in the application cease to 
be under the BSD L?  I would argue "no" if I don't own the copyrights to 
the BSDL code in the first place (any modifications I make might be 
another matter however).  Does this prevent the work as a whole from 
being licensed by me under the GPL including all components?  Not by any 
reasonable reading I can give.

Similarly, does the GPL work as a whole clause prevent me from using GPL 
components, creating a work, releasing that work under a dual-license 
model (standard GPL or at the option of the customer a standard EULA but 
with the addition of warranty terms)?  If the user always has the choice 
to use the GPL terms, I would argue that the license has not been violated.

In short, is there anything in the MS-PL that precludes use of the code 
from complying with all of the other terms set in the GPL v3?  I don't 
see any.
>
>
>>> Your interpretation of the word "by" seems to be equivalent to 
>>> "which is accomplished by".
>> No actually I don't.  I see it as requiring that MS-PL licensed areas 
>> to the extent that they are identifiable must be licensed under the 
>> terms of that license.  This means including the license and 
>> (probably, so as not to run amok with GPL code copyright owners) 
>> identifying the portions of the code with reasonable legal notices 
>> such as:
>>
>> /* The block of code including the below function is licensed under 
>> the MS-PL license.  Please see accompanying ms-pl.txt for details, 
>> copyright (c) [yyyy] [author_name] */
>
> But is this licensed only under MS-PL, as required by MS-PL, or is it 
> also licensed under GPL v3, as required by GPL v3?
>
Unfortunately by your reading, BSDL code would be incompatible to (new 
or old) because under copyright law, you can only distribute it under of 
the terms granted by the copyright holder anyway(!) but IANAL.


Does New BSDL == GPL?  Does copying BSDL code into a GPL v3 app 
automatically transfer permissions sufficient to solve this problem 
(since you can only distribute *any* copyrighted work under permissions 
set forth by the *copyright owner*)?

In that this clause of the MS-PL does nothing outside of summarize what 
copyright law in general says (again IANAL), I fail to see how this is a 
problem by itself.

Best Wishes,
Chris Travers


["text/x-vcard" not shown]