Github user hvanhovell commented on the issue:
https://github.com/apache/spark/pull/21580
LGTM. It is better UX to have a more descriptive error messages.
However I do like the idea of being able to use window functions in
filters. I often use the following pattern:
```scala
val df: Dataframe = ...
df.select(row_number().over(Window.partitionBy($"key").orderBy($"seq")).as("rn"))
.filter($"rn" === 1)
.drop("rn")
```
Teradata, for example, has the `qualify` filter clause for these cases.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
