Invariant Properties

I’ve worked with proto-GIS software at a number of positions – that is, I had dynamic data and background information (roads, urban areas, etc.), all in long-lat coordinates. But it was the dark ages of computer science and we didn’t have access to free high-quality libraries so it was all roll-your-own. Fortunately in both cases (meteorology and aviation) we just needed the ability to display static and very constrained dynamic data, albeit sometimes in interesting ways. (E.g., creating vertical of the atmosphere along great circle lines.)

This meant I was really looking forward to PostGIS in Action from Manning Press. I’ll admit that a lot of that is morbid curiosity – how long would it take to do a task today vs. the late 90s?

The first step is to get the software. Most of it is already packaged in Ubuntu packages for postgresql-8.4, postfix, and postgresql-8.4-postfix. Unfortunately the standard packages do not include shp2pgsql-gui and we must build it ourselves.

The process is relatively straightforward if you’re comfortable building C projects. The GUI client is in the loader directory and you’ll need to specify the path to the header files and libraries in your Makefile. My changes are listed below.

# GTK includes and libraries
GTK_CFLAGS = -I/usr/include/gtk-2.0 -I/usr/include/glib-2.0 -I/usr/lib/x86_64-linux-gnu/glib-2.0/include -I/usr/include/cairo -I/usr/include/pango-1.0 -I/usr/include/gdk-pixbuf-2.0 -I/usr/include/atk-1.0 -I/usr/lib/gtk-2.0/include
GTK_LIBS = -lgtk-x11-2.0

Finally copy the executable to /usr/local/bin.

I now create the postgis template and test database according to instructions in Appendix B.

$ psql (as postgresql superuser)
CREATE DATABASE template_postgis WITH TEMPLATE = template1 ENCODING = 'UTF8';
\c template_postgis;
CREATE LANGUAGE plpgsql; --this may not be needed if running 8.4
\i /usr/share/postgresql/8.4/contrib/postgis-1.5/postgis.sql;
\i /usr/share/postgresql/8.4/contrib/postgis-1.5/spatial_ref_sys.sql;
\i /usr/share/postgresql/8.4/contrib/postgis_comments.sql;
UPDATE pg_database SET datistemplate = TRUE WHERE datname = 'template_postgis';
GRANT ALL ON geometry_columns TO PUBLIC;
GRANT ALL ON spatial_ref_sys TO PUBLIC;

CREATE DATABASE testgis WITH TEMPLATE = template_postgis;
\q

and download and install a number of shape files downloaded the sites mentioned in the back of the book.

At this point I have a GIS-enabled database containing a few gigabytes of data…. now what?

That’s a good question. So far I’ve had a few hours to play with two Ubuntu packages: OPENJump and qgis.  The latter can read files from PostGIS and it looks like it will be easy to do full relational queries on the data. That is, do things like “show the location of all McDonalds within 50 miles of a wilderness area.” OPENJump also has a ‘query’ function but I don’t know how powerful it is.

Still… it makes me wonder about alternative approaches to those earlier problems. E.g., what would be the costs and benefits of breaking the old presentation architecture into two pieces. The first would take the scientific data and convert it to GIS XYM (and XYZM) data and the second would be a standard GIS client. It adds a step and it might complicate things like animations… but it would also make it trivial to provide the information in multiple forms since there would be complete decoupling of the science and the presentation.