[GRASS-dev] GRASS 7 Migration from CVS to SVN

[Hamish]

[combining some posts here]

Jan:
> sorry for the late coment. I am catching up the list only in certain
> intervals and usually can not react immediately unless put in cc
> or notified else.

Ok. I hope you saw all the comments from a number of us about how happy we
have been with the CVS and other support from Intevation. That is
probably the number one reason we have not moved to OSGeo's servers a year
ago and why this discussion continues on and on.

> this list apparenly is driven by the wish to go to osgeo.

I would say there is some general wish to have closer ties with OSGeo than
we have now. There is no reason to move away from Intevation, just a
slight reason to put something at OSGeo. In that case, all other things
being neutral, and having to do a cvs2svn port anyway, a move to OSGeo has
had some support.
(my personal view is that I'd be happy with either place)

Markus:
> > We clearly see that GForge for GRASS is way under-used.
Jan-Oliver Wagner:
> I guess you mean the bug tracker is underused?

yes. (AFAIK that's all we have turned on)



I think it is best to solidly decide which support software we want to
use. Once we have clearly decided that we just push hard to get it ;)
Right now we seem to be deciding what to use (& where) based on what is
already installed, which seems backwards.

What support software to use seems to be a totally separate issue from
where to host things, as Intevation seems to be happy to provide what we
ask for as best they can, and I guess (but don't know) that the OSGeo
admins would be supportive too. For me, getting the support service
software "right" is much more important than looking like a good OSGeo
team member or keeping *everything* on the same server in some unified
content management system. But of course we should encourage tie-ins with
OSGeo as best we can, I don't mean to say we shouldn't do that as much as
possible, just not at the expense of choking the good infrastructure we
have now.


For my 2c, I like:
* current web server setup seems ok (including rsync mirrors and CVS)
* current wiki software (MediaWiki) (we can fix the current spam issue)
* move to SVN for source code management
  (actually, for me anyway, I don't have any problems with CVS)
* Same ViewCVS setup as we have now, but interfaced to SVN:
  e.g. http://josef.fsv.cvut.cz/cgi-bin/viewcvs.cgi/?root=grass7svn
* .... bug tracker .... what was wrong with the old one again?
 The two really annoying things with Gforge bug tracker to me is the
 top posting of comments and that you can't cc it in posts to the
 grass-dev mailing list, you have to use the web form, duplicating
 effort and killing casual contributions. I don't like to think that
 we are defeated in what we want to do based on spammers.


I don't see a huge need to move Everything into "the" OSGeo framework with
a OSGeo "look" to it. On the same lines I don't see a huge need to combine
all web/wiki/SCM/bugs under a single Trac or Gforge master. (I accept
Frank's word that the SVN tie-in will be rewarding, but haven't seen that
clearly yet myself) I accept that there can be e.g. multiple SVN web
interfaces running at the same time, to suit all tastes -- at least for
that it is important to note that it doesn't have to be mutually
exclusive. Another point is that the CMS (trac,gforge,..) doesn't have to
actually host the wiki, e.g., it just has to point to where it is.

So what I really care about is having the right tools working at 100%, not
where they live at night or what colors they are painted. If we decide to
stay with GForge or MediaWiki but there is some feature we just can't live
without, then surely there is enough programming talent around here to
patch them & submit to upstream..... and if Intevation or OSGeo needs
help fulfilling our wants, we can do that too.


regards & sorry for the ramble,
Hamish