This is the mail archive of the newlib@sourceware.org mailing list for the newlib project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: CodeSourcery's licensing of newlib files


Being pedantic here but sourceware.org is just a host
that (I recall) Redhat has provided for years. It is
not a corporate entity or non-profit by itself -- just
a domain for development.

The company who did this is CodeSourcery which is
now part of Mentor Graphics.

In general, my opinion is that they have been
good at upstreaming patches so I would think the
technical part of the patch against the base newlib
they started with is small. With the disclaimer that
sometimes ports like the NIOS II linger in their code
a long time before submitted.

It is a shame that no one from there has commented.

Did they do this to the source for gcc, binutils, etc?

On 10/24/2013 10:06 AM, Jonathan S. Shapiro wrote:
> On Thu, Oct 24, 2013 at 12:43 AM, Ralf Corsepius
> <ralf.corsepius@rtems.org>wrote:
> 
>> On 10/23/2013 10:09 PM, Gregory Fong wrote:
>>
>>> On Wed, Oct 23, 2013 at 6:03 AM, freddie_chopin <freddie_chopin@op.pl>
>>> wrote:
>>>
>>>>  Can they attach sth like that to newlib's files?
>>>>
>>> Yes, that's permitted under the BSD license.
>>>
>> Correct, but it means their changes can not be applied to sourceware.org's
>> newlib.
> 
> 
> If by "they" you mean "SourceWare", then that's incorrect. SourceWare can
> sign appropriate releases and donate their code.
> 
> If by "they" you mean "SourceWare customers", that's a bit more nuanced.
> 
> If I build a patch against the SourceWare tree, and it can be applied
> successfully to the newlib tree without change, I'm probably fine. The
> patch itself is my property, not SourceWare's. It is not a derived work,
> and therefore not subject to their copyright. The fact that it can be
> applied successfully to the baseline newlib tree confirms that my patch has
> no proprietary SourceWare content in it. That means that (a) I'm not
> violating the SourceWare copyright, which only applies to the SourceWare
> content, and (b) No confidential or trade secret information exists in my
> patch (because the patch applied to baseline newlib successfully).
> 
> If the patch *won't* apply to baseline newlib, then things get interesting.
> At that point I think you need a lawyer to figure it out.
> 
> Practically speaking, the impediment here isn't an intellectual property
> issue. It's the fact that the user has to migrate/convert the patch to
> apply against the newlib sources before submitting.
> 
> 
> Aside: if we knew the baseline newlib version that SourceWare had
> appropriated, it would be really interesting to know (a) the number of
> lines in the diff between SourceWare and baseline, and (b) how many files
> actually changed. While the BSD license *does* allow SourceWare to add the
> notices, you can only assert copyright on material that you own. If
> SourceWare has systematically altered the copyright notice on files they
> did not change, then (a) that's fraud, and (b) it would make it very
> difficult for them to enforce successfully. It's obviously not a battle
> anyone wants to have, and the newlib team is right to be careful about
> incoming patches, but I'd hazard to guess that SourceWare has overstepped
> very badly here.
> 
> To say that more plainly: BSD allows you to redistribute code under a broad
> range of license terms. BSD does *not* grant you any right to claim
> ownership of work you did not create. A copyright notice is a claim of
> ownership.
> 
> 
> Jonathan S. Shapiro
> 



-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel.sherrill@OARcorp.com        On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
Support Available                (256) 722-9985


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]