SQL: Can a single OVER clause support multiple window functions?

Question

Let's say I have a list of events related to viewing pages, each of which has a session identifier. For each event, I want to add the time and URL of the chronologically first pageview in this event session. For example, suppose my events are in a table testthat looks like this:

uid | session_id | timestamp | url
----------------------------------------------------
u1    0            0           a.com/
u1    1            1           a.com/p1
u1    1            2           a.com/p2

I need a SQL command that creates the following:

uid | session_id | timestamp | url      | s_timestamp | s_url
---------------------------------------------------------------------
u1    0            0           a.com/     0             a.com/
u1    1            1           a.com/p1   1             a.com/p1
u1    1            2           a.com/p2   1             a.com/p1

The window functions seem to be here, but I'm new to them. The following statement creates the desired table, but I wonder if it is suboptimal

SELECT
    uid,
    session_id,
    timestamp,
    url,
    first_value(url) OVER (PARTITION BY uid, session_id ORDER BY timestamp ASC) s_url,
    first_value(timestamp) OVER (PARTITION BY uid, session_id ORDER BY timestamp ASC) s_timestamp
FROM test

, , OVER. URL- , OVER? SPARK SQL, , SQL.

Answer 1

WINDOW, , ​​ SELECT:

SELECT
    uid,
    session_id,
    timestamp,
    url,
    first_value(url) OVER w s_url,
    first_value(timestamp) OVER w s_timestamp
FROM test
WINDOW w AS (PARTITION BY uid, session_id ORDER BY timestamp ASC)

Apache Spark SQL HiveQL.