[
https://issues.apache.org/jira/browse/SPARK-17262?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15748307#comment-15748307
]
Michel Lemay commented on SPARK-17262:
--------------------------------------
I agree. As a realworld example of this, consider KMeans:
It has a structure KMeansSummary, which in turn holds a transient reference on
the Dataframe it used to fit the model.
SizeEstimator count the size of this Dataframe but is clearly not related to
the model itself in memory. Technically speaking, a KMeans should only have
it's clusters centers and some metadata.
> Spark SizeEstimator does not ignore transient fields in java classes when
> calculates class size
> -----------------------------------------------------------------------------------------------
>
> Key: SPARK-17262
> URL: https://issues.apache.org/jira/browse/SPARK-17262
> Project: Spark
> Issue Type: Bug
> Components: Spark Core
> Affects Versions: 1.6.2, 2.0.0
> Reporter: George Shuklin
> Priority: Minor
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
