Libgfortran licensing

Libgfortran licensing

[Tobias Schlüter]

To make things short: under which license should libgfortran be licensed?

While doing some cleanups on libgfortran I noticed that some files are
licensed under the plain GPL, others under the LGPL, I didn't see any that say
GPL+exception, but there could well be. Could someone who has the authority to
make such statements for the FSF please tell me under which license
libgfortran should be licensed (I assume that the library should be licensed
under the terms of the GPL+exception, as GCC's other runtime libraries).

Also, some files say that they're part of libgfor others say libgfortran. I
assume these should all say libgfortran, right?

Regards,
- Tobias Schlüter

Re: Libgfortran licensing

[Mike Stump]

On Aug 31, 2004, at 1:20 PM, Tobias Schlüter wrote:
> To make things short: under which license should libgfortran be 
> licensed?

libgcc2.c, affectionately known as GPL+exception at least for all 
standard runtime components.

Re: Libgfortran licensing

[Toon Moene]

Tobias Schlüter wrote:

> To make things short: under which license should libgfortran be licensed?
> 
> While doing some cleanups on libgfortran I noticed that some files are
> licensed under the plain GPL, others under the LGPL, I didn't see any that say
> GPL+exception, but there could well be. Could someone who has the authority to
> make such statements for the FSF please tell me under which license
> libgfortran should be licensed (I assume that the library should be licensed
> under the terms of the GPL+exception, as GCC's other runtime libraries).

Stop !   Wait !

Only the original author(s) of these pieces can determine under what 
license they want their contributions distributed.  As "we" didn't write 
all of them (some were written by Andy Vaught), we cannot decide 
unilaterally to change the licensing.  If the license does not conform 
to the one normally used for run-time library code, and we didn't write 
the specific code ourselves, we should recode the functionality.

-- 
Toon Moene - e-mail: toon@moene.indiv.nluug.nl - phone: +31 346 214290
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
A maintainer of GNU Fortran 95: http://gcc.gnu.org/fortran/

Re: Libgfortran licensing

[Robert Dewar]

Toon Moene wrote:

> Stop !   Wait !
> 
> Only the original author(s) of these pieces can determine under what 
> license they want their contributions distributed.

Is that true? I thought that the copyright holder, in this case the
FSF, can make that determination. I don't remember the assignment
having a limitation on using licenses MORE liberal than the GPL.

Re: Libgfortran licensing

[Paul Brook]

On Tuesday 31 August 2004 23:02, Toon Moene wrote:
> Tobias Schlüter wrote:
> > To make things short: under which license should libgfortran be licensed?
> >
> > While doing some cleanups on libgfortran I noticed that some files are
> > licensed under the plain GPL, others under the LGPL, I didn't see any
> > that say GPL+exception, but there could well be. Could someone who has
> > the authority to make such statements for the FSF please tell me under
> > which license libgfortran should be licensed (I assume that the library
> > should be licensed under the terms of the GPL+exception, as GCC's other
> > runtime libraries).
>
> Stop !   Wait !
>
> Only the original author(s) of these pieces can determine under what
> license they want their contributions distributed.  As "we" didn't write
> all of them (some were written by Andy Vaught), we cannot decide
> unilaterally to change the licensing.  If the license does not conform
> to the one normally used for run-time library code, and we didn't write
> the specific code ourselves, we should recode the functionality.

Doesn't the copyright assignment give the FSF power licence the code as they 
see fit?

IIRC libiberty only had problems changing the licence because the code wasn't 
FSF copyright.

Paul

Re: Libgfortran licensing

[Robert Dewar]

Paul Brook wrote:

> Doesn't the copyright assignment give the FSF power licence the code as they 
> see fit?

I don't think so, I think the assignment document guarantees that the
licensing will be compatible with the GPL. You don't want the FSF to
suddenly be able to make everything proprietary (the original author
has this capability, but it is deliberately signed away with the
assignment).
> 
> IIRC libiberty only had problems changing the licence because the code wasn't 
> FSF copyright.
> 
> Paul

Re: Libgfortran licensing

[Toon Moene]

Paul Brook wrote:

> On Tuesday 31 August 2004 23:02, I wrote:

>>Stop !   Wait !
>>
>>Only the original author(s) of these pieces can determine under what
>>license they want their contributions distributed.  As "we" didn't write
>>all of them (some were written by Andy Vaught), we cannot decide
>>unilaterally to change the licensing.  If the license does not conform
>>to the one normally used for run-time library code, and we didn't write
>>the specific code ourselves, we should recode the functionality.

> Doesn't the copyright assignment give the FSF power licence the code as they 
> see fit?

Yep, *if* we can determine that everything "we", i.e., you and Steve, 
brought over to the gcc-g95 sourceforge repository at the time of the 
fork was written by people who assigned their copyrights to the FSF.

As far as I know it's a non-trivial exercise to check this, although 
Andy did, of course.

-- 
Toon Moene - e-mail: toon@moene.indiv.nluug.nl - phone: +31 346 214290
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
Maintainer, GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
A maintainer of GNU Fortran 95: http://gcc.gnu.org/fortran/

Re: Libgfortran licensing

[Tobias Schlüter]

Paul Brook wrote:
> Doesn't the copyright assignment give the FSF power licence the code as they 
> see fit?
> 

Yes, with one restriction, it also says:
"4. FSF agrees that all distribution of the Works, ..., shall be on terms that
 ... permit anyone ... to redistribute copies ..." (some legalese and closer
specification left out)

- Tobi

Re: Libgfortran licensing

[Mike Stump]

On Aug 31, 2004, at 3:02 PM, Toon Moene wrote:
> Only the original author(s) of these pieces can determine under what 
> license they want their contributions distributed.  As "we" didn't 
> write all of them (some were written by Andy Vaught), we cannot decide 
> unilaterally to change the licensing.  If the license does not conform 
> to the one normally used for run-time library code, and we didn't 
> write the specific code ourselves, we should recode the functionality.

The purpose of assigning copyright to the FSF is to allow the FSF to 
correct goofs like this.  If that was done for Fortran, and it should 
have been, then the FSF can decide to change it.  The right global 
rights person that would review this _knows_ if they can without asking 
the FSF for permission, and _will_ ask them otherwise.

So, short story long, it is ok to do up the patch and submit it, but it 
_must_ be reviewed by the right person.  Law in the past has had some 
state on this topic, and might be the right person, the SC can ok it as 
well.  Not that they can, but rather, they know if they can, and won't 
if they can't.

All the libfortran maintainers are supposed to know all of this...  :-) 
  so it isn't meant to controversial.  In addition, the library 
reviewers really should catch this sort of thing _before_ oking work, 
not after.  :-(  It is always more painful after.

Re: Libgfortran licensing

[Kai Henningsen]

dewar@gnat.com (Robert Dewar)  wrote on 31.08.04 in <4134F8A8.70301@gnat.com>:

> Paul Brook wrote:
>
> > Doesn't the copyright assignment give the FSF power licence the code as
> > they see fit?
>
> I don't think so, I think the assignment document guarantees that the
> licensing will be compatible with the GPL.

It doesn't, which is why we have a GFDL (which is certainly not GPL- 
compatible) and why I have no assignment.

> You don't want the FSF to
> suddenly be able to make everything proprietary

Well, not proprietary, but they've already demonstrated that relicensing  
under non-DFSG-free terms is possible - see the GFDL debate.

> (the original author
> has this capability, but it is deliberately signed away with the
> assignment).

It's strictly the copyright holder who gets to decide this.

> > IIRC libiberty only had problems changing the licence because the code
> > wasn't FSF copyright.

Whereas my memory says the FSF (that is to say, rms) simply didn't want  
to.

MfG Kai

Re: Libgfortran licensing

[Robert Dewar]

Kai Henningsen wrote:
> dewar@gnat.com (Robert Dewar)  wrote on 31.08.04 in <4134F8A8.70301@gnat.com>:
> 
> 
>>Paul Brook wrote:
>>
>>
>>>Doesn't the copyright assignment give the FSF power licence the code as
>>>they see fit?
>>
>>I don't think so, I think the assignment document guarantees that the
>>licensing will be compatible with the GPL.
> 
> 
> It doesn't

That's incorrect, see later quote in this thread
> 
> It's strictly the copyright holder who gets to decide this.

Nope, it's subject to contract like anything else