[Proj] "Double Ellipsoid" error, reproduction

[Frank Warmerdam]

Gerald I. Evenden wrote:
> I find this thread so incredible confusing I have tried to stay out of it 
> entirely but I have one, maybe two questions:
 >
> I thought the basic detail operations of datum conversion were done in 
> geographic data space (latitude-longitude) or perhaps x-y-z.  That is, you 
> have to have the data in geographic space to do the datum conversion 
> calculation.   True or False?
> 
> Also, ellipsoid factors are not part of the datum conversion as long as the 
> data is in geographic coordinates:  True or False?
> 
> If the above is true, what does the ellipsoid values have to do with anything?  
> That is a rhetorical question and the answer is obviously: nothing.

Gerald,

There are several methods for datum shifts.  Grid shift files seem to be
normally applied as latitude and longitude offsets varying over the region
of the grid shift file.  But the "towgs84" 3 and 7 parameter shifts are
applied in geocentric space.  So the lat/long values need to be converted
into geocentric coordinates in the source datum/ellipsoid, modified, then
converted from geocentric into lat/long again in the destination
datum/ellipsoid.  Improper ellipsoid values results in potentially
substantial differences.

> If the issue is simply two completely separate operations--- datum conversion 
> and ellipsoid-projection--- then what is all the discussion about???

This is why I believe it was suggested we should be able (potentially)
to use different ellipsoid values for the projection calculations and
for the datum shift.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent