[Q-e-developers] gitlab

[Stefano Baroni]

Thank you, Michael: you got the point. 
Let me try to summarise the main points of the lengthy discussion that has been going on over the past few months and of the decision that we are slowing making, trying to reach as general a consensus as possible.
The software stack running qe-forge is obsolete and there is no viable path to upgrading it. A decision has been firmly made to dismiss qe-forge and to move to a service based on git.
We know that many present and prospective users are already registered on GitHub and we understand that GitHub would be the preferred choice for them. We note however that the underlying software (git) is the same and that GitLab can be accessed using a GitHub account.
This being said, concerns have been raised about the policy of GitHub regarding the property of the metadata and the location of the data on servers outside the UE (it seems that this may be an issue in the application to future project calls). It was decided therefore that it would be wise to either have the project hosted in a private instance of whatever service we choose, or on a public service that would make the migration to a private instance as seamless as possible in the (near) future.
As I understand, private instances of GitHub are exceedingly expensive. At the same time, it seems that none of the partners of the QEF is at present ready to host and administer a private instance of a git repository.
It was therefore (tentatively, but almost definitively) decided to:
Move the project to gitlab.com <http://gitlab.com/>, where all the development effort will take place.
Mirror the repository to a private GitLab instance, which will likely be hosted by CINECA, thus ensuring that all the data are under our complete control in EU.
Look forward to moving the development to the private repository if and when the QEF or some of its partners can afford the move.
Have mailing lists managed as a separate service, likely but not necessarily, by the center that will host the private gitlab instance.
The current QE mirror on GitHub will continue to exist and to be updated.
The move will take place in the next few days. Paolo says that we might consider to move at revision 14000 (we are 8 commits away).

Hope to have got all the main points right and to have reported them faithfully.

Stefano

> On 10 Nov 2017, at 21:27, Michael Banck <mbanck at gmx.net> wrote:
> 
> On Fri, Nov 10, 2017 at 07:40:41PM +0100, Nicola Marzari wrote:
>> On 10/11/2017 11:04, Samuel Poncé wrote:
>>> Dear all,
>>> 
>>> My guess is that the main reason is that GitLab is Ukrainian vs US for 
>>> GitHub and that GitLab is free and open source.
>>> 
>>> Its has slower access time for me through (sometimes takes 10 solid 
>>> seconds before getting the webpage).
>>> 
>>> When is the official move planned (i.e. commit to SVN are not accepted 
>>> anymore) ?
>>> 
>>> Best,
>>> Samuel
>> 
>> github is also free, and with private repositories, if you ask and
>> mention the educational status. anyhow, great to move to git - I just
>> think we should also have considered that many in the community
>> are already familiar and registered with github.
> 
> The main difference is that you can host gitlab yourself (if you want),
> though nowadays that might be somewhat involved due to the Docker
> integration and whatnot.
> 
> If you plan to use a hosted service, there's probably not a huge
> difference between github and gitlab so going where most of the
> developers already are makes sense.
> 
> The main issue might be that neither provide some mailing-list like
> discussion forum, as the forge does/did.
> 
> 
> Michael



—
Stefano Baroni -  SISSA, Trieste - http://stefano.baroni.me, stefanobaroni (skype)

There are two ways of doing a theoretical calculation: you should have either a clear physical model in mind, or a rigorous mathematical basis. You have neither. [E. Fermi to F.J. Dyson, as humbly reported by the latter]

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20171121/3fe35543/attachment.html>