Doubles are equivalent for our purposes and work better with ExpressionEncoders (BigDecimals were being converted to null when serialized into Rows as part of #106). Spark JTS geometry constructions are happy with either, so this shouldn't affect anything.
Double
s are equivalent for our purposes and work better withExpressionEncoder
s (BigDecimal
s were being converted tonull
when serialized intoRow
s as part of #106). Spark JTS geometry constructions are happy with either, so this shouldn't affect anything.