[Qgis-psc] Moving tickets to GitHub - status update?

Richard Duivenvoorde rdmailings at duif.net
Tue Jun 21 03:19:15 PDT 2016


On 20-06-16 17:59, Jürgen E. Fischer wrote:
> I guess it's more the hosting and not redmine itself that makes it slow - and I
> believe that's the main point why we consider to switch.

Yes, slowness, BUT also people able/willing to invest time into
maintaining a(n updated) new Redmine instance. Except you, Sandro,
Pirmin, Alex and a handfull of others nobody stepped up to really fix it.

I tried to contact a commercial Redmine provider and asked to
fix/migrate for us, but they only migrate to their commercial Redmine
service.

> And I doubt tags only are better suited to classify a massive amount of tickets
> like ours than what redmine offers.  But I have neither experience with using
> redmine for anything else but QGIS and no experience with using gh with many
> tickets.

I googled some big Github repo's:

https://github.com/twbs/bootstrap/issues (>12000 issues)

https://github.com/rails/rails/issues (>8000 issues)

So yes, you are right. On the other hand, if I'm correct Mapserver used
the move to github to start more or less with a clean sheet...

> Matthias' script IMHO looses too much information we have in rm - and trying to
> do more - back in April - I quickly ran into gh's api throtteling.   IIRC there
> was also didn't find a way to attribute stuff to the original
> submitter/commenter.  

Could this be overcome by splitting the work in some days: first only
the last(first?) 1000 issues, next day... etc etc?

Original submitter, as long as we have the id/name of him/her (as text)
somewhere in the issue, we know who to attribute/contact if really needed?

> That also points to the fact that gh is proprietary service - which is another
> thing I have issues with - IMHO a route we should be careful about.  For git
> that doesn't matter - but for issues (and that also goes for PRs) it does.

I'm with you about proprietary services. But because Github is so big
(too fail), I think most other services at least have an import route
for it IF we want to leave again.

And looking at the discussions now: while Redmine is 'open source' we
are also more or less 'trapped' into it.

Using whatever system on a scale like QGIS does, there will be problems
migrating.

> gh is of course much easier for us to use - because we don't have to worry
> about hosting, scalabilility and administration - which one of gh's big
> plus.   Integration is another - but that cuts both ways.

I think the hosting, scalability and administration are the main points
to move. BUT also the almost defacto standard of even a beginning
developer starts by creating a github account and repo :-)

More 'open' options like: Gitlab or Gog I think miss the 'general
familiarity' for normal people/users.

In last PSC meeting we talked about 'working groups', if all
(complaining) people in this thread invest time/money into the move, I
think a 'Move to Github'-working group would be a nice first temporary
'working group'?
Maybe schedule an online "GH-migrating hacking weekend/day"?

Regards,

Richard Duivenvoorde

ps: I think the mantra thing was ok to work with, as I have seen
OSGEO-SAC fighting wiki spam like beasts! And without the mantra, spam
bots kept popping up all around us...



More information about the Qgis-psc mailing list