Re: [DebianGIS] [OSGeo-Discuss] packaging FOSS GIS for Ubuntu in education
Ok, following-up very fastly, a few ideas could be extended and
On Sun, Oct 28, 2007 at 12:14:34PM -0400, Frank Warmerdam wrote:
> 2) Review what is missing from this in DebianGIS and try to find volunteers
> to help the DebianGIS project package the appropriate software. There is
> a fair amount of expertise needed for Debian packaging (IMHO), but a few
> volunteers willing to invest 30-50 hours over the coming 6-7 months could
> make quite a difference. But we need to realize DebianGIS is a well
> established project with it's own culture and expectations and be prepared
> to work within this.
That would be particularly useful for the d-gis project, of course.
> It might be helpful for OSGeo to maintain a Debian system (as a VM or a
> whole blade) to provide a working and testing environment for folks who
> don't run Debian at home/work.
I think it would be a nice thing maintaining something like
the mapsnack virtual appliance (http://mapsnack.sourceforge.net/).
I see that at FOSS4GIS2006 and it is based on Debian+vmware. A VMware image
like that would be useful in order to provide a complete up-to-date and
stable-based distribution, with a few useful backported packages for GIS key applications.
That would be useful for users of course. Other VMs would be nice
but much more complicated to be used and provided, and currently
VMplayer can be used without license fees to run this sort of images.
A development environment would be based essentially on sid, but it
could be installed as a 'dchroot' environment and installed on a etch
workstation, too. Using an amd64 box one could provide both
i386/amd64 dchroots for lenny, etch and sid, with just minimal
maitainance. I'm not persuaded that providing a blade facility
would be better than having the same environments included in a VM image
for interested developers anyway, but for free CPU time. I'm thinking
essentially to something like the Qtopia development environmenti for
those people who know that...
I'm available to provide details and possibly setup a framework on
a VM image to be used as a model for an hosting facility, etc. I would
separate anyway profiles of use (users vs developers/packagers)...
> 3) Find out what is needed to bring UbuntuGIS up to the appropriate
> I don't know what people are involved in UbuntuGIS or how they operate. My
> understanding is that for major new Ubuntu releases the UbuntuGIS packages
> are built from the DebianGIS source packages.
AFAIK currently ubuntugis is largerly on the track of DebianGis, as
simple derivation (by rebuilding and minimal patching).
I know some secondary packages in universe should be integrated in
debiangis, anyway (after checking their level of quality and status).
> 4) Putting this together on CD/DVD is where the rubber hits the road. It
> would be great if Venka can lead this aspect, but I'm sure he would
> appreciate help. There is no reason that a first draft of this can't
> be prepared based on existing packages.
> One cool things is that Debian, and regular network based Ubuntu uses also
> all benefit from the upstream efforts. I love this sort of leverage!
> I'm not sure what OSGeo can do to facilitate this activity. We obviously
> can't direct volunteers, only encourage them. We don't have funding
> targeted for such an effort. However, if a modest amount of money
> could make a big difference I might be able to scare some up.
> I'm confident we can provide mailing lists, wiki space, server space,
> and bandwidth if these are helpful.
> I have cc:ed Frankie, the DebianGIS lead, in the hopes he could comment
> on how we can help support the DebianGIS effort.
As someone could know we already have our internal facilities for
maintainance. They are not easly exportable to other platforms (non
Debian-based) unfortunately because it would imply a giant job
of ad hoc development and more work for maintaince of the facilities.
Of course directly contributing to DebianGis would be the most straight
way to maximize contribution effectiveness. I think another viable
platform for those things would be gentoo-based (as already done for
Francesco P. Lovergine