[ 
https://issues.apache.org/jira/browse/SPARK-61498?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Hyukjin Kwon updated SPARK-61498:
---------------------------------
    Fix Version/s: connect-rust-4.3.0

> Raise the errors of the conf read of the origin of an expression, and end the 
> wait of an execution at the second Ctrl-C while its interrupt hangs, in the 
> Python client as PySpark does
> ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: SPARK-61498
>                 URL: https://issues.apache.org/jira/browse/SPARK-61498
>             Project: Spark
>          Issue Type: Bug
>          Components: Connect
>    Affects Versions: connect-rust-4.3.0
>            Reporter: Hyukjin Kwon
>            Priority: Major
>             Fix For: connect-rust-4.3.0
>
>
> Follow-ups of a post-hoc review of SPARK-61470, checked against PySpark 4.3.0:
> - The origin of an expression (F.col, the Column methods, groupBy and the 
> other functions that PySpark wraps with _with_origin) reads 
> spark.python.sql.dataFrameDebugging.enabled once, through 
> is_debugging_enabled. The Python client dropped the errors of that read. So a 
> Ctrl-C during the first F.col of a session against a server that does not 
> answer was lost: the call returned, and every later origin call waited again. 
> PySpark's _with_origin lets the error raise: KeyboardInterrupt at the first 
> Ctrl-C, or the error of the server.
> - During an execution whose interrupt hangs or is retried (UNAVAILABLE), as 
> against a server that does not answer it, the second Ctrl-C aborted the wait, 
> but the wait then awaited the interrupt. So only a third Ctrl-C ended it, 
> where PySpark stops at the second.
> - Column.__str__ no longer captures an origin, as PySpark's Column has no 
> __str__ of its own for _with_origin to wrap.



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to