[Qgis-psc] Migration to Github

[Denis Rouzaud]

Le jeu. 15 févr. 2018 à 13:03, Jürgen E. Fischer <jef at norbit.de> a écrit :

> Hi Denis,
>
> On Thu, 15. Feb 2018 at 15:05:14 +0000, Denis Rouzaud wrote:
> > On practical points, proof of concept was achieved by Matthias (and some
> > others) and showed all of this is possible except attributing comments to
> > the original author...bummer ;).
>
> News to me.  The migration didn't go though - I thought Matthias had given
> up
> on the migration, because he was uncertain that we'd go through with it
> and it
> might end up as a waste of energy.  And his target was not to do a full
> migration - IIRC.
>

Well, it depends on what we mean by migration of course (moving all
tickets? all tags? etc).

>
> I tried a full migration - and ran into obstacles from github (eg.
> throtteling)
>

proposition was to contact Github, correct?


> and had doubts about mapping all what we have in redmine to tags in github


well, again depends on what we're migrating. History or tags. My
proposition is to start with a new empty list for QGIS 3 and migrate issues
on demand (with an assitant tool, triggered from Redmine or semi-manually).
And for the history, to keep Redmine as read-only.


> -
> and how useable it would be which such a large number of issues and/or
> tags as
> ours.
>
> Richard tried again later and also gave up.
>
> Attributing comments to the original author was just one thing that
> couldn't be
> done at all.
>
> We finally settled to upgrade and move redmine to a faster machine - and
> that
> was that.
>

Upgrading Redmine should certainely help the migration as the original
migration tool was for more recent version, correct?

>
> And when this was brought up again, we didn't talk about the implementation
> (ie. how we map the trac/redmine ticket numbers to github issues - our
> commit
> messages reference the tickets).
>
> And then there was this sudden vote...
>

I understand that. Let's wait for Madeira then.


>
> > (because, damn, it's not integrated with the source code!).
>
> Of course it is linked to the source code.  It's just not linked to github,
> because github doesn't allow us to - otherwise we'd just address PRs and
> issues
> differently and let one point to github and the other to redmine.
>

Of course, I meant source code repo!

I think we can discuss for ever on what has been done and what is possible.
Then, we can adapt the decision tree and add the question:

If we can't keep the whole history, shall we stick to Redmine?
And what do we mean with whole history (added files, comments, tags, etc).

Denis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-psc/attachments/20180215/1695d2ca/attachment.html>