Re: [Opengrads-devel] Shapefile extension: first crack

[Arlindo da S. <da...@al...>]

On 11/27/07, Efren Serra <efr...@me...> wrote:
> http://mapserver.gis.umn.edu/docs/howto/svg-howto
>
> Note that Mapserver is built on the GDAL/OGR library and hence, it can be
> extended to pretty much generate anything one wants from a single ESRI
> Shapefile.

Actually, I was asking whether SVG would be a viable INPUT format for
Mapsever. I guess I need to understand what exactly you would like to
use GrADS for, and then find the best way to interface to Mapserver.
Let me digress just bit...

1) As you know GrADS graphics are all vector instructions. At this
point even fonts are rendered internally. Image generation relies on
libraries such as gd (printim) and cairo (gxyat) for rasterization.

2) At a higher level GrADS can produce contour plots, filled contours
(shaded) plots, streamlines, wind barbs, wind vectors, station model
plots (with rendered). An these can be produced without any maps,
frames, axis, etc. Just naked plots at the surface of the earth. Are
these the kind of output that you would like to ingest into Mapserver
for further compositing?

3) Right now these can be written out as vector output in encapsulated
postscript, PDF or SVG. In the beginning of this thread I commented
that one might be able to write these vector instructions as
shapefiles (.shp, .shx, and .dbf) --- I need to read the shapefile
white paper to better understand the functionn of these 3 files.

4) If SVG is a viable *input* format for Mapserver then a little tweak
in the on-line version of gxyat might allow us to write some extra
metadata to make Mapserver happy. This could be a quicker approach to
achieve what you need, bypassing shapefiles altogether. Not that
shapefiles wouldn't be useful for truly GIS desktopers...

5) Here at NASA/GSFC we provide our homegrown WMS server with nice PNG
images which it then servers to the world, e.g.,

               http://www.map.nasa.gov/cgi-bin/tc4-d5fcst-hwl.cgi

I am assuming that you are pursuing a vector graphics approach for
speed and scalability. Am I correct?

      Arlindo