Re: Announcing GNOME's official GitHub mirror

[fr33domlover <fr33domlover mailoo org>]

No problems, GNOME having read-only mirrors can be useful to people.

Just make sure there's an easy way to opt out. For example, I wouldn't
want any of my code automatically uploaded to GitHub. I think every
maintainer should have the right to cancel mirroring for their module.

If GitHub was free software, decentralized, etc, then I could maybe
agree that mirroring can be activated by default for existing and new
modules. But considering the nature of GitHub, I consider it somewhat
rude to mirror a module without letting a maintainer an option to cancel
it, or make it disabled by default and allowing the maintainer to switch
it on.

On ה', 2013-08-15 at 13:20 +0100, Emmanuele Bassi wrote:
> hi Luis;
>
> thanks for answering.
>
> On 15 August 2013 13:00, Luis Menina <liberforce freeside fr> wrote:
> > Le 15/08/2013 12:44, Emmanuele Bassi a écrit :
> > > > Actually, the fact that we have to ask to opt out is an issue in
> > > > itself. We shouldn't even have to. This should have been opt in from
> > > > the start. People (maintainers and commiters in this case) shouldn't
> > > > have to fight to get back what you have taken away from them.
> > >
> > > considering that this is a mirroring system of a distributed version
> > > control system, I'm puzzled as to what has been lost. you still have
> > > all your rights to the software you maintain and commit to, and you
> > > still have the right to push your work to more than one repository.
> > > care to elaborate a bit more on this?
> >
> > I'm not a maintainer, but it seems to me that a maintainer may want as
> > few entry points for patches as possible, or at least not need to poll
> > to find patches. We already have bugzilla, or git.gnome.org. If extra
> > clones exist and seem officially endorsed by GNOME, and there's no
> > process to send those patches upstream, this clearly means it's up to
> > the maintainer to poll for patches on these extra clones.
>
> as I said the last time the idea of a github clone was being floated
> around, I don't want to look in multiple places for patches. nor I
> want to get pull requests from mirrors I don't maintain directly — and
> even then, I basically always say that if a patch is not on Bugzilla,
> then it doesn't exist.
>
> the work that Alberto did, though, seem to be clear that: a) the
> canonical place for submitting patches is Bugzilla, and b) the GitHub
> clones are for mirroring only, so that people can easily create a
> public fork on their own GitHub account when they wish to hack on
> something. it is, essentially, a read-only mirror. as a maintainer, I
> don't have a problem with exposing my code on multiple venues — that's
> what I do already every day.
>
> ciao,
>  Emmanuele.