Re: Repos

[Ivan Vučica]

It may be theoretical, until the bnetd-type event comes along.

Regarding Git, I don't value ordered numbers that much. They're not all that much more important than hashes, and tracking where code got forked and merged seems more useful to me in a distributed environment. Git's CLI is highly annoying to me, but I can live with it.

Things that seem core benefits for project such as GNUstep:

- Git values smaller, more easily reviewable patches, while still letting them being sent in 'all at once'. 

- Git allows easier merging in such cases, while still staying 'up to date' in a fork.

- Git more easily survives messing around with metadata, if absolutely needed. That means we can more easily add tools around it.

Just an example of tools we could add: a web interface for code review integrated with a mailing list. Multiple people could review code, any contributor could sign-off on it, but at least one must sign-off on it.

I don't think such a code reviewing tool would necessarily work out for GNUstep, but with SVN, we can't even think about it.

On Sat Dec 21 2013 at 11:06:36 AM, Dr. H. Nikolaus Schaller <hns@goldelico.com> wrote:

> Am 21.12.2013 um 11:40 schrieb David Chisnall:
>
> >
> > On 21 Dec 2013, at 10:05, Dr. H. Nikolaus Schaller <hns@goldelico.com> wrote:
> >
> >> Hm. I start to wonder why is the copyright assignment needed at all to get code into GNUstep?
> >
> > The FSF requires copyright assignment for two reasons.  The first is that they may wish to change the license.
>
> Well, why should that be done?
>
> >  We've done this in GNUstep in the past, moving some GPL'd code to LGPL when it was moved from an application into a library (this would be simplified by making all of the GNUstep tools LGPLv2.1 or later, which would also simplify distribution, but that's another issue).
>
> Ok, I see - but how often does it happen? And in that case (by using the signed-off:) it is not too difficult to identify the authors.
>
> >
> > The more important issue is that only the copyright holder has standing to sue for copyright infringement.  If someone uses GNUstep code in violation of the license, we'd probably like the FSF to chase them.  They can't do this nearly as easily if the copyright is held by a group of individuals.  There's also the related issue that if someone reaches a settlement with the FSF no one else can sue them over the same bit of code.  This is a problem for Linux, because a few kernel devs are somewhat obsessive about the GPL and have a habit of suing companies over GPL violations, but even if you settle with them there's no guarantee that no other kernel devs will sue you.  The FSF, as part of a settlement, sells a retroactive commercial license for FSF software, so at the end of losing a case against them you have the the license that you need and no one else can sue you.  This is a big stick that helps them in negotiation.
>
> Ok  I see. But isn't that theoretical? How many such cases did exist in the past 10 years? And how many patches and developers did we loose because of missing assignment?
>
> > So, while copyright assignment is annoying, it's likely required as long as GNUstep is distributed under a restrictive license.
>
> I always thought that GPL is not restrictive... Except one thing that nobody can change the licence of someone else's code.
>
> > It's less important for the runtime, which is MIT licensed, because you have to try really hard to violate the MIT license...
>
> Ok, I see.
>
> Nikolaus
>
>
> _______________________________________________
> Discuss-gnustep mailing list
> Discuss-gnustep@gnu.org
> https://lists.gnu.org/mailman/listinfo/discuss-gnustep