GNAT runtime licensing mess

GNAT runtime licensing mess

[Patrick Noffke]

This is prompted after I received a quote from AdaCore for the "safety-critical" GNAT development tools and runtime.  Let me preface this by saying I have great respect for AdaCore and I want them to continue to be successful.  But, to summarize, I think they ought to focus on selling support and tools, not messing with the runtime license.

I noticed many source files in the GNAT-GPL runtime (gnat-gpl-2014-47 for ARM-ELF) differ only from the FSF version (specifically libgnat-devel-4.9.2-6.fc21.x86_64 on Fedora 21) by omitting this clause in the comment header of the GNAT-GPL version of the file:

-- As a special exception under Section 7 of GPL version 3, you are granted --
-- additional permissions described in the GCC Runtime Library Exception,   --
-- version 3.1, as published by the Free Software Foundation.               --

This got me thinking -- who are AdaCore to choose whether or not to include this clause?  Some files are copyrighted by AdaCore, some by the FSF, and some older copyrights by Universidad Politecnica de Madrid and The European Space Agency.  In addition, some files credit the GNAT team at New York University and some of the GNARL files credit the GNARL team at Florida State University as well as the Technical University of Madrid.

Some files in the zfp-support-gpl-2014 runtime still include the runtime library exception (particularly some .S and .ld files -- likely files missed by some tool used to remove the exception from *.ad?).

At some point "the GCC developers" (whatever that means) decided/agreed to include the FSF's runtime library exception into GCC.  How they pulled that off seems like magic -- how do they know they got every author to agree to that exception?

But now AdaCore, while certainly a significant contributor to the runtime but not the sole author, gets to decide to include or exclude the exception based on whether or not you pay them?  What's to stop anyone else from becoming an author of the runtime and putting that exception in files that don't have them?

I get the spirit of the GPL, but the FSF has spoken about why GCC has the exception:  https://www.gnu.org/licenses/gcc-exception-faq.html

I get that, too.  So given that, who really has the right to change the license of any particular GNAT runtime?

And this is of practical concern.  I have added support for the TM4C123 MCU in the zfp-support-gpl-2014 runtime.  I'm quite happy to transfer the copyright to the FSF (though I'm not sure how to go about that).  But going forward, if I include the runtime exception, does that mean AdaCore simply cannot remove it from the files I wrote?  What should I do to "give back" my work?  It seems very confusing and messy.

Of course, I am neither seeking nor offering legal advice, rather asking for your perspective or informed opinion.

Best regards,
Patrick

P.S.  Right after where the exception is/should be is this clause:

-- You should have received a copy of the GNU General Public License and    --
-- a copy of the GCC Runtime Library Exception along with this program;     --
-- see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see    --
-- <http://www.gnu.org/licenses/>.                                          --

I could not find a copy the GCC Runtime Library Exception in the GNAT-GPL version.

Re: GNAT runtime licensing mess

[Florian Weimer]

* Patrick Noffke:

> This got me thinking -- who are AdaCore to choose whether or not to
> include this clause?

This is explicitly permitted under the terms of the GNU General Public
License, version 3, in section 7: “When you convey a copy of a covered
work, you may at your option remove any additional permissions from
that copy, or from any part of it.”  (The GCC run-time library
exception uses the additional permissions framework established by the
GPL.)

Re: GNAT runtime licensing mess

[David Botton]

Much has been said on the topic, I'd look at previous threads over the last 10 years. But since important that no one ever think people the Ada community supports such self serving small actions:

1. The basic answer is as Patrick wrote, they have permission to remove the exceptions, but you can opt to use the FSF versions and not promote the libre site, etc. http://www.getadanow.com  - Perhaps people could send me links and I'd even host or link to other platform versions there as well.

2. What they have done and are doing has been a low blow to the community and is disrespectful and condescending to other contributors of open source Ada software and for the work of others they built their own work on. No one action has sent away or prevented growth in Ada use more (in my opinion, and is perhaps just my own)

3. Despite the above, they have and so far continue to contribute their work back to the FSF so if your work helps them to continue development and sell their products as long as their derivative work is still flowing back to the FSF I wouldn't be concerned at all.


David Botton

Re: GNAT runtime licensing mess

[Simon Wright]

David Botton <david@botton.com> writes:

> Despite the above, they [AdaCore] have and so far continue to
> contribute their work back to the FSF so if your work helps them to
> continue development and sell their products as long as their
> derivative work is still flowing back to the FSF I wouldn't be
> concerned at all.

This doesn't appear to be true for the bare-board RTS versions.

Re: GNAT runtime licensing mess

[David Botton]

> This doesn't appear to be true for the bare-board RTS versions.

How long has there been bare-board versions out? Is that the case for gcc 5.0 as well?

David Botton

Re: GNAT runtime licensing mess

[Patrick Noffke]

On Tuesday, February 24, 2015 at 11:59:18 AM UTC-6, Florian Weimer wrote:
> * Patrick Noffke:
> 
> > This got me thinking -- who are AdaCore to choose whether or not to
> > include this clause?
> 
> This is explicitly permitted under the terms of the GNU General Public
> License, version 3, in section 7: "When you convey a copy of a covered
> work, you may at your option remove any additional permissions from
> that copy, or from any part of it."  (The GCC run-time library
> exception uses the additional permissions framework established by the
> GPL.)

Thank you for that.  It answers part of my questions.

Patrick

Re: GNAT runtime licensing mess

[Simon Wright]

David Botton <david@botton.com> writes:

>> This doesn't appear to be true for the bare-board RTS versions.
>
> How long has there been bare-board versions out? Is that the case for
> gcc 5.0 as well?

No System.BB in GCC 5:

--  This is the root package of the bare board implementation of the low level
--  tasking support for Ravenscar compliant run times.

Re: GNAT runtime licensing mess

[Waldek Hebisch]

Patrick Noffke <patrick.noffke@gmail.com> wrote:
> I noticed many source files in the GNAT-GPL runtime (gnat-gpl-2014-47
> for ARM-ELF) differ only from the FSF version (specifically
> libgnat-devel-4.9.2-6.fc21.x86_64 on Fedora 21) by omitting this clause
> in the comment header of the GNAT-GPL version of the file:
> 
> -- As a special exception under Section 7 of GPL version 3, you are granted --
> -- additional permissions described in the GCC Runtime Library Exception,   --
> -- version 3.1, as published by the Free Software Foundation.               --
> 
> This got me thinking -- who are AdaCore to choose whether or not to include this clause?  
<snip>
> At some point "the GCC developers" (whatever that means) decided/agreed
> to include the FSF's runtime library exception into GCC.  How they
> pulled that off seems like magic -- how do they know they got every
> author to agree to that exception?

As Florian wrote anybody can drop exception.  FSF insists that
copyright of any code included in GCC is transfered to FSF.  As
a copyright owner thay can add extra permissions.

-- 
                              Waldek Hebisch
hebisch@math.uni.wroc.pl