Off-by-one in Robinson projection?

[QuLogic]

In Robinson, i is limited by:
/~https://github.com/OSGeo/proj.4/blob/330e8bf686f9c4524075ca1ff50cbca6c9e091da/src/PJ_robin.c#L87-L91
to 0 <= i <= NODES - 1 and i + 1 is never used in s_forward. Here, NODES is set to 18, but there are 19 entries in X and Y. In s_inverse, i + 1 is used.

So, is the last entry in the X and Y tables supposed to be applied in s_forward (and accidentally not), or is it only there for s_inverse? If this is intentional, I guess a comment would be helpful to make this clear. Of am I just misreading something here?

[kbevers]

I have no idea, to be honest. It does seem plausible that there's been made a mistake here. Usually when these types of problems show up I have a look in the original papers that describe the projection. I haven't got the time for that at the moment so I would suggest you do that yourself if you need a quick answer to this :-)