[GRASS5] GRASS and lat/long with sphere

[Glynn Clements]

Markus Neteler wrote:

> > while trying to define a lat/long location with Sphere instead
> > of WGS84, I found in
> > 
> > src/libes/gis/ellipse.table
> > # earth as a sphere:
> > # There is a bug in libes/gis/get_ellipse.c, where a function
> > # return e=0.0 in any case. So do not use sphere!
> > # sphere        "Spherical Earth"               a=6370997.000
> > 
> > Is there no/any hope to fix the code (I don't know how...).
> 
> [Note that also for Mars planet GIS activities the sphere is
>  required. So we should really support it]
> 
> ... of course I tried it by activating the "sphere" line. What happens:
> 
> - select lat/long
> 
> Please enter a one line description for location <sphere>
>  
> > sphere test
> =====================================================
> sphere test
> =====================================================
> ok? (y/n) [y]
> WARNING: Line 144 of ellipsoid table file
> 
> </mpa_sw/ssi/BIO/software/GRASS5.0.0/grass5bin_cvs/grass5/etc/ellipse.table>
> is invalid
>  
> Please specify ellipsoid name
> Enter 'list' for the list of available ellipsoids
> Hit RETURN to cancel request
> >
> 
> The warning occurs at
> src/libes/gis/get_ellipse.c
> line 329

The function get_a_e2_f() in get_ellipse.c requires that e2 (the
eccentricity-squared) is greater than zero (otherwise it returns zero,
indicating an error).

Changing it to allow e2 to be equal to zero doesn't result in any
problems which are immediately obvious. However, it may be that some
code can't cope with this (e.g. computing the reciprocal). In all
probability, the only way to find out is to make the change and wait
for bug reports.

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