PostgreSQL and PostGIS offer to create indexes over GiST with pretty much the same code. On 2D items, "This gives R-tree behavior" is what it says in the PostgreSQL source comments. However, it is a bit different than the original R-Tree, an observation being that the split technique is the Korotkov split. Furthermore, on PostGIS source, there are comments suggesting R*-tree ideas. I am somewhat unsure…
So my question is: is the "R-Tree behaviour" offered in PostgreSQL the vanilla R-Tree? If not, what are some of the differences? Is this data loaded insert by insert or is there a bulk loading mechanism in place?
Best Answer
I'm not sure what is the vanilla R-Tree but there is internal citations inside the source file you linked. Also, the R-Tree inside of PostGIS is directly copied from PostgreSQL with these minor modifications (as in the FAQ),