Hi, No, you can't. If you use ScalaUdf, you can like this;
val bv = sc.broadcast(100)
val testUdf = udf { (i: Long) => i + bv.value }
spark.range(10).select(testUdf('id)).show
// maropu
On Sun, Dec 18, 2016 at 12:24 AM, 李斌松 <libinsong1...@gmail.com> wrote:
> The spark hive udf can read broadcast the variables?
>
--
---
Takeshi Yamamuro
