Github user rednaxelafx commented on a diff in the pull request:
https://github.com/apache/spark/pull/19083#discussion_r135985618
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/codegen/CodeGenerator.scala
---
@@ -1001,6 +1001,16 @@ abstract class CodeGenerator[InType <: AnyRef,
OutType <: AnyRef] extends Loggin
}
object CodeGenerator extends Logging {
+
+ // This is the value of `HugeMethodLimit` in JVM settings. Since we
can't get this value on
+ // runtime and this is fixed in released JVMs, we hard-code this values
here:
+ //
+ // $ java -XX:+UnlockDiagnosticVMOptions -XX:HugeMethodLimit=99999
+ // Error: VM option 'HugeMethodLimit' is develop and is available only
in debug version of VM.
+ // Error: Could not create the Java Virtual Machine.
+ // Error: A fatal exception has occurred. Program will exit.
+ private val hugeMethodLimit = 8000
--- End diff --
Actually this threshold is only meaningful on HotSpot VM and some
HotSpot-derived JVMs. Other JVMs, for instance IBM J9 doesn't use the same
threshold. It'd be a bit too strict and biased to make this a non-configurable
behavior for all JVMs.
I'd suggest that if we are to do this, at least centralize the JVM
detection logic somewhere (e.g. unify with the JVM detection logic in
`org.apache.spark.util.SizeEstimator`) and only set this kind of threshold
based on the detected JVM. That way we're much less likely to regress on other
JVMs, and/or even on future versions of the HotSpot VM where it'll get a new
JIT compiler (Graal) with new JIT compilation heuristics that could be
different from the current version.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
