[PROJ] Relevant descrepancies between Sinusoidal and Homolosine

[Even Rouault]

On mardi 10 mars 2020 14:49:37 CET Luí­s Moreira de Sousa wrote:
> Hi again Even, apologies for insisting. In the .ods I sent earlier you can
> verify that the Homolosine is preserving the northing-latitude identity,
> even with the ellipsoid. And also that it does not match the Sinusoidal.
> 
> My guess is that the Homolosine is already spherical-spherical as it is.

Ah indeed... This P->es = 0 thing wasn't used for the instanciation of the 
sinusoidal, so yes my PR was completely useless. Sorry for the noise.

So, igh preserves the identity because it does force es = 0 to the underlying 
sinu... Whereas sinusoidal standalone will use an ellipsoidal formulation for 
an ellipsoid.

As far as whether sinusoidal for the ellipsoidal case is correctly implemented 
or not, I've no idea. I'm not sure your CSV file clearly demonstrates the loss 
of the equal-area property(at least, this is not obvious to me), since it 
probably compares spherical vs ellipsoidal formulations of different 
projections.

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com