[
https://issues.apache.org/jira/browse/JEXL-146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14154717#comment-14154717
]
Henri Biestro commented on JEXL-146:
------------------------------------
An Interpreter instance is created for each expression/script execution; it
should not be possible to share an interpreter between 2 threads so the thread
safety (of isStrictEngine) is not really a concern. The only exception is the
'cancelled' flag which allows thread cancellation of a script running as a
callable.
You are however absolutely right about the potential issue with the functors;
these should definitely not be shared between interpreter instances. I've made
corrections accordingly; Closure/callable no longer acquire a reference to
their caller functor. Thanks for spotting the issue.
> Performance problem in Interpreter.unknownVariable mechanism
> ------------------------------------------------------------
>
> Key: JEXL-146
> URL: https://issues.apache.org/jira/browse/JEXL-146
> Project: Commons JEXL
> Issue Type: Bug
> Affects Versions: 2.1.1, 3.0
> Reporter: David Maplesden
>
> When JEXL is used in non-strict mode it silently ignores unknown variables.
> However it still constructs a {{JexlException.Variable}} exception, even
> though it doesn't throw it.
> Constructing exceptions in Java is expensive. If you happen to have a
> situation where unknown variables are common then this becomes a performance
> problem. It is showing up as an issue in our application.
> Inspecting the source of {{Interpreter}} it looks like a similar issue could
> also occur with a number of the other exceptions. They are always created,
> even if they are not going to be thrown.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
