[GRASS-dev] GRASS-TNG

[Roger Bivand]

On Thu, 22 Feb 2007, Hruby Martin wrote:

> Hello all,
> 
> I am Radek's supervisor in his master thesis which should be a part of 
> this mentioned project. The project itself seems to be too huge to become 
> an one-man-show. For this reason, we have started asking our friends and 
> collegues in our country, like Martin Landa and Jachym Cepicky for their 
> help and comments.
> 
> Opening idea is that GRASS is too old fashioned to keep it in development 
> in current programming style. Some people will agree, some people not. We 
> just want to remake the GIS kernel and to rewrite current modules to the 
> new fashion. The whole philosophy of GRASS should remain the same.

Which was why I wrote to the original questioner, suggesting that trying
to implement a small sample of GRASS modules on top of Terralib, which is
a full, bells&whistles templated C++ LGPL GIS framework written by some of
the best people around in GIScience (Gilberto Câmara and others), would be
a worthwhile test. If (some of)  the legacy GRASS library infrastructure
can be replaced by a different kernel, and run acceptably, then we would
know something. 

My guess would be that the MAPSET concept would need time, probably also
the LOCATION concept, but that other things, like locking, would be
handled by the underlying database systems. Raster reading would also need
fine tuning. The vector modules might largely prove unneccesary, because
Terralib has good vector support.

Not as brave a a whole new kernel, just using an existing new kernel that 
meets your design criteria, for what they are worth. It is also feasible 
in an MSc, which writing a real modern GIS kernel isn't. Knowing som 
Portuguese might help!

Once you have some running and profiled prototypes, there may be more 
interest, otherwise I doubt it, to be honest. If the prototypes run with 
the same options and output effects as the originals, the GUI will have a 
chance because they are (to a large extent) scripted on top of the CLI 
modules.

Roger

> 
> So, this is a fork to main stream development. Or, better say, this is 
> more a research. If someone wants to join, then is welcome ;-) This 
> project may be a very nice experiment.
> 
> regards,
> 
> Martin Hruby
> 
> 
> On Thu, 22 Feb 2007, Michael Tiemann wrote:
> 
> > On Thu, 2007-02-22 at 09:15 +0100, Radek Bartoò wrote:
> >> On Thursday 22 of February 2007 08:33:12 Paolo Cavallini wrote:
> >>> So why not rewriting NVIZ from scratch (perhaps usin the paraview
> >>> approach)? *This* would be more or less feasible in one year, it would
> >>> not impact on other grass modules, and it would be *much* needed and
> >>> appreciated by the GRASS community, I think.
> >>> pc
> >>>
> >>> Radek Bartoò ha scritto:
> >>>> On Wednesday 21 of February 2007 18:05:51 Trevor Wiens wrote:
> >>>
> >>
> >> Actually this kind of project was one of I thought about first since I'm more
> >> oriented to CG and terrain visualization than GIS. It shouldn't be a NVIZ
> >> replacement but 3D real-time visualization tool reading data from GRASS.
> >
> > I'll demonstrate my ignorance and suggest that blender/verse may be such
> > a tool waiting to become part of the suite: http://www.uni-verse.org/
> >
> > I spoke with the verse developers at SIGGRAPH last year, and they
> > explained that the reason they make their channel information (such as
> > RGB or Z or whatever) all 64-bits is to support scientific
> > visualization.
> >
> > M
> >
> > _______________________________________________
> > grass-dev mailing list
> > grass-dev at grass.itc.it
> > http://grass.itc.it/mailman/listinfo/grass-dev
> >

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no