[GRASS5] Database linking question
Eric G . Miller
egm2 at jps.net
Wed Jan 3 00:33:15 EST 2001
On Tue, Jan 02, 2001 at 07:57:21PM -0800, Rich Shepard wrote:
> I'm not a GIS data storage guru (nor do I want to be one). But, postgres
> handles graphic objects so I wonder why topology functions couldn't be added
> to the base that already exists. After all, it's advertised as an
> "object-relational dbms", and there are built-in graphics data types.
> Even when data are stored as arc-node sets, couldn't the connections be
> stored in an associated table? I, much less than anyone here, do not
> understand the issues involved.
Well, I'm no guru either (though I play one on mailinglists ;).
Anyway, I actually worked with the guy trying to do FMaps (see
sourceforge). I did come up with a few different ideas about
representing spatial data in PostgreSQL (or any RDBMS actually).
Extending the type system was one approach, though it's hard to do
topological data that way. The other approach, which seems more
general, would be to use a catalog system and a bunch of tables to
represent the levels of abstraction. The latter system really needs
some kind of "middleware" to manage the abstraction, but in theory
should be possible in any relational database. I did a little playing
with schemas to support such a thing and my preliminary investigations
showed promising results with the structure. Unfortunately, I was a bit
overworked at the time and didn't have the energy to pursue it.
Then I accidentally started hacking on GRASS code...
Actually, I think it's quite possible to manage all of the data that
something like GRASS manages directly in a relational database. It
might even be more efficient if designed properly. I don't see GRASS
moving in that direction though, at least not without a fundamental
redefinition of what GRASS is and how it manages data.
One of these days, I'll get back to working on this stuff.
Unfortunately, I see massive demands on my hacking energy from my job
for the next six months (makes it hard to come home and hack away on
another project). I'm hoping Frank Warderdam will pull a rabbit out
of his hat with this OSVecDB thing (though, if mailinglist traffic is
any indication, it won't be anytime soon...).
Anyway, that's enough ramblin'.
Eric G. Miller <egm2 at jps.net>
If you want to unsubscribe from GRASS Development Team mailing list write to:
minordomo at geog.uni-hannover.de with
subject 'unsubscribe grass5'
More information about the grass-dev