[
https://issues.apache.org/jira/browse/SPARK-37055?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17432159#comment-17432159
]
Hyukjin Kwon commented on SPARK-37055:
--------------------------------------
Actually, it's more to prevent running Spark jobs only for the sake of input
validation. For example, assume a pandas API requires to have the same values
in its input:
{code}
def abc(df):
if self.sort_values() != df.sort_values()
raise Exception("all values have to be same")
{code}
and assume that the input {{df}} contains a very complicated computation chain.
For example:
{code}
df = spark.read.csv().sort().repartition().sort().agg(...)
{code}
{code}
another_df.abc(df) # would result in computing `df` two times (+ sort each df).
{code}
So, this JIRA aims to have the eager check by default (to match with pandas'
behaviour) but provide an option to avoid such expensive computation.
> Apply 'compute.eager_check' across all the codebase
> ---------------------------------------------------
>
> Key: SPARK-37055
> URL: https://issues.apache.org/jira/browse/SPARK-37055
> Project: Spark
> Issue Type: Umbrella
> Components: PySpark
> Affects Versions: 3.3.0
> Reporter: dch nguyen
> Priority: Major
>
> As [~hyukjin.kwon] guide
> 1 Make every input validation like this covered by the new configuration.
> For example:
> {code:python}
> - a == b
> + def eager_check(f): # Utility function
> + return not config.compute.eager_check and f()
> +
> + eager_check(lambda: a == b)
> {code}
> 2 We should check if the output makes sense although the behaviour is not
> matched with pandas'. If the output does not make sense, we shouldn't cover
> it with this configuration.
> 3 Make this configuration enabled by default so we match the behaviour to
> pandas' by default.
>
> We have to make sure listing which API is affected in the description of
> 'compute.eager_check'
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
