[OSGeo-Discuss] OSGeo guidelines for code hosting ?

[Mateusz Loskot]

On 17 October 2015 at 21:34, Anita Graser <anitagraser at gmx.at> wrote:
> On Sat, Oct 17, 2015 at 7:59 PM, Sandro Santilli <strk at keybit.net> wrote:
>> On Sat, Oct 17, 2015 at 05:06:54PM +0200, Anita Graser wrote:
>> > On Oct 17, 2015 12:10 PM, "Mateusz Loskot" <mateusz at loskot.net> wrote:
>> >
>> > What about the projects which are already on Github?
>>
>> They hadn't wasted any effort, because no effort existed...
>
> Projects which moved to Github at some point spent effort to both make the
> decision and make the move. Granted, no effort on OSGeo side might have been
> wasted.

Let's imagine, all projects move to GitHub, then it affects OSGeo infrastructure
which becomes obsolete. Also, for some/many, move to GitHub may be
controversial, depending on actual (F)OSS-orientation of the community members.
So, IMO, wider discussion is not that pointless.

>> Anyway, the request is to the OSGeo board to clarify requisites of hosting
>> for OSGeo projects.
>
> I'm trying to understand the direction this discussion is going since I
> might not be aware of the full background story. Are you expressing a wish
> for stricter hosting guidelines? Or would you simply want clarification,
> even if that means OSGeo takes a very "everything goes" position?

I'd like to clarify my part of the discussion, which began as informal
chat on IRC,
since Sandro called my name, here is the story:

1. PostGIS is considering switch to Git, so they started discussing
git.osgeo.org vs GitHub.
2. Sandro prefers the idea of git.osgeo.org, so he started working
with SAC towards
setting up git.osgeo.org hosting, Trac integration, etc.
3. Regardless, PostGIS team is leaning towards GitHub.
4. Some of OSGeo projects have already moved to GitHub.
5. I started asking if Sandro's & SAC efforts to set up git.osgeo.org
make any sense,
if it is not going to be wasted energy - in case OSGeo projects stick
to SVN or move to GitHub.
6. Sandro does not mind and has been pressing on
7. I pointed out, that it is not about wasting energy of an individual
volunteer, there is more to consider,
    especially, if all stars on the skyp indicate so far, that just
PostGIS (what is still unsure)
    is going to use git.osgeo.org

Once Sandro has completed setting up git.osgeo.org infrastructure, who
is going to maintain it?
If SAC is going to inherit that baby, has SAC agreed to take it over
and invest time on keeping it up
and running? It is going to be one (or more) services extra to keep up
to date, secure, monitor,
back up, migrate, etc.

Bottom-up approach is a fantastic thing, but it drags certain
consequences that need to be considered.

So, I proposed that we (SAC) should survey the Community about their
hosting needs and ask
if git.osgeo.org is something we actually need, if there are any
projects interested in actually using it.

If yes, do we need plain Git installation, Git+Trac or perhaps GitLab?
If not, then why?
If not, because projects prefer GitHub, BitBucket, etc. then perhaps
OSGeo should
explicitly allow that and, perhaps, let SAC to actively maintain
organization accounts
on those external hosting services.

Unless, we as an established organization do not Wild Wild West approach,
we may need to clarify some answers.

Finally, after reading about similar experiences [1] at
Eclipse/Apache, I think this issue
is related to the recent "OSGeo is becoming irrelevant" thread too, so
it may be something
to make the Board concerned about.

[1] http://redmonk.com/sogrady/2013/06/20/eclipse-github/

I hope it makes more sense now.

Best regards,
-- 
Mateusz  Loskot, http://mateusz.loskot.net