[PROJ] Do you use Chebyshev polynomial coeffiecient approximations in proj?

[Even Rouault]

> Projections
> aren’t that expensive to compute by modern standards so there is
> simply no need to approximate them.

Just to react on this, and not on the Chebyshev specific case, coordinate 
transformation can be a significant CPU eater. For example in the case of 
gdalwarp, a piece-wise affine approximation is used by default to save CPU 
cycles and avoid doing coordinate transformation for potentially billion of 
pixels.

> So I’d like to remove the
> functionality. There’s no need to waste ressources on maintaining
> code and documentation that in all likelihood is not used anymore.
> 
> Before I go ahead and remove the code I want to check with the list
> if anyone still has a need for this? If no-one responds positively to that
> I am going to remove the code at the end of this week.

No objection from me. Never used it. Didn't know that it existed.

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