Hi all,
While I am generally happy with RGeoS as it is now (using version 9.0.7) I think what it is missing, is an easy to use possibility to limit the area of kriging to within the outline of the sampling pattern. It is to some extent possible to achieve that effect with a moving neighbourhood, but I think implementing a convex hull function (as in sp) would be worthwhile as it would decouple the definition of the area to be estimated from the neighbourhood settings, which would give greater flexibility.
Regards,
Hannes Höffle