[mapserver-users] Strategy for dealing with Census Data?

[Walt Lin]

> 1. Where is the "tgr2shp" binary/source?  I don't seem to have
> that utility in my collection.  Please tell me it's not an Arcview
> extension. ;^)

It's a $99 program from www.gistools.com (it only runs on Windows and is
slow, and can only process so many zip files at a time).

> 2. Is it true Census 2000 data is freely available in .shp format?
> I see tiger data for sale but haven't found FREE census 2000
> shapefiles.

Back in June I could not find it anywhere in .shp format-- I don't know
about know, but I doubt the entire set is freely available in shp
anywhere (hence the need for a tgr2shp program).

> 3.  Finally, strategically,  is there a performance trade-off by
> using 1 large shapefile versus several shapefiles?  For example, a

Of course.  Mapserver will have to load one huge file and look for
intersecting points/lines/etc instead of loading smaller smaller smaller
smaller smaller smaller smaller smaller smaller files.  In
fact, I even added another layer of abstraction by storing the paths to
shp files in Postgres with an R-tree index-- to process a map request I
query Postgres for files that intersect the box, and then tell mapserver
to load only those files when drawing the map.  It actually works very
well, I think.  You can also TILE your shapefiles.