On 4 Feb 2017, at 07:24, Fred Kiefer <address@hidden> wrote:
> No reason to apologise. It was great to bring this issue to our attention. Something must be done and I agree we should rather act early than late. I would suggest you add the source to your local gitlib instance and from there to Savannah and GitHub and set up the automatic synchronisation mechanism. We call the Savannah instance the main one, but advice people to get data from GitHub. That way we get the best of both worlds and everybody should be happy.
By far the biggest advantage of GitHub is the ability to accept pull requests (well, that and the fact that most people already have a GitHub account, so can file bug reports with a very low barrier to entry). If the GitHub repository is not the authoritative one, then you need to ensure that you have bidirectional sync, so pushes to the GitHub repo are mirrored in the official one. It’s possible to do this via the hooks that they provide, but it’s potentially racy and if two people push to the GitHub and other repos at the same time then sorting out the history can be nontrivial.
We moved libobjc2 to GitHub a while ago and I’ve seen a noticeable improvement in contributions and useful bug reports. I haven’t seen anything similar for the rest of GNUstep. I don’t intend to get into a rehash of the last debate, but once again GNUstep has the choice of either making contributions easy and encouraging wider community participation or doing something ideologically pure. My vote is for the former option.
One of the biggest problems we face is the fact that we are a gnu project and, thus, have to make sure that people know we can't accept unassigned contributions.
There have been a few pull requests on github that we couldn't accept or had to go through a long process because of this.
Many people on this list WILL disagree with me on this and admonish me that assignments are absolutely positively needed and necessary and they are more "comfortable" being part of gnu.
I care about the project itself. Political bs is not useful to me. I care about the software. While I tangentially agree with RMS' purpose and vision I don't agree with the methods we often use to get there.
I would very much like to leave GNU in order to get rid of the copyright assignment requirement so that we could accept range of contributions.
I have no illusions, however, that the contributions will come pouring in once we open the gates.