Feedback on sterea / gauss projection implementation.

[SamMousa]

I have implemented the sterea projection:
https://github.com/SamMousa/Proj4/blob/master/src/Projection/Sterea.php

However I'm not getting correct results when comparing to proj4 (the C variant).

Could you check if you see any obvious errors?

[SamMousa]

Okay I've got it working (took the implementations they used in proj4php) and am getting close to the same result as cs2cs:

Mine: [6.6028105991312, 53.294205897322]
Cs2cs: [6.60652472415, 53.2969942548]

This probably has to do with the fact that after doing the inverse sterea projection I still need to convert to the proper ellipsoid (my source EPSG:28992 uses bessel).

Note that my implementation doesn't "integrate" well now, since I've chosen to do the projections as simple classes. This means that they take all the parameters they need in the constructor; for Gauss for example you need the eccentricity squared ($es) of the ellipsoid, so you must pass that in. For Sterea, which extends Gauss the length of the semi-axis A is required, so to construct Sterea you must pass in both.

This gets pretty verbose but guarantees clear dependencies. For the projection the datum for example is not used at all, so passing in a coordinate (which has datum and projection information included) is a bit overkill.

Of course when we validate this approach it then becomes easy to build 1 or more helper classes that allow functionality similar to commandline tools like cs2cs, just passing in data in 1 format and getting it out in another.