Vector (2d) Fubar in 40beta20

when trying to sift a non-integer value that has at a previous time come out of a Vector (2d Join) node (even if the nodes are not directly connected), Sift (Value) fails to find the value for an epsilon <= .1^9

(i am really curious about the source of that one. what has floating point precision to do with the vector?)

sifttest5.v4p (9.6 kB)

it is a vector thing.

houston.v4p (3.8 kB)

push… no text …

nice find. fixed for beta>21.
that should spare us quite some singledouble conversions…