[GRASS5] bug: d.vect.area and HTMLMAP

[Glynn Clements]

Markus Neteler wrote:

> sorry for a new bug report: It seems that d.vect.area does not work properly
> with the HTMLMAP driver (d.area was fine).
> 
> Please update:
>  cvs up src/scripts/contrib/grassmirrorsmap
> then 
>  cd src/scripts/contrib/grassmirrorsmap
>  mkgrassmap.sh
> 
> The resulting 'grass_sites.gif' is fine, but the 'grass_sites.htmlmap' does
> not contain the expected polygon definitions. Using the old d.area the
> procedure went fine.
> 
> I need that to generate the mirrors world map:
> http://grass.itc.it/index.html
> 
> Eric, could you please check for a difference between the old and the
> new implementation which may concern the HTMLMAP driver?

The HTMLMAP driver only implements two drawing operations, R_text()
and R_polygon_abs(). d.area uses R_polygon_abs(), while d.vect.area
uses G_plot_area().

Eric added G_plot_area() for this purpose. It is based upon
G_plot_polygon(), but it handles areas containing holes. Both of these
functions "scan convert" the shape, drawing a series of horizontal
lines using application-defined functions. For d.vect.area, the
functions are D_move_abs() and D_cont_abs().

The bottom line is that the HTMLMAP driver can't realistically
"reverse" the scan conversion to obtain the original polygons (the
original shape may not even be a polygon, in the usual sense of the
word), and re-writing d.vect.area to use R_polygon_abs() isn't
feasible (that would involve a tesellation algorithm, which is far
from trivial).

So, it seems that we will need to maintain d.area for use with the
HTMLMAP driver. I'll re-enable it in src/CMD/lists/GRASS.

-- 
Glynn Clements <glynn.clements at virgin.net>