[
https://issues.apache.org/jira/browse/SOLR-16707?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17703209#comment-17703209
]
Michael Gibney commented on SOLR-16707:
---------------------------------------
{quote}
I could imagine timeAllowed support being added for async. That needn't be a
blocker for enabling this by default, though.
{quote}
Agreed, timeAllowed support could be added for async. I don't see how that
affects the question at hand though. {{async}} is already currently enabled by
default. {{async=false}} is not required for anything. {{async=true}} generally
performs better, is required for a number of features, and adds a trivial
amount of memory overhead (24 bytes per cache entry). So the question is:
should we remove support for {{async=false}}? And if not, should we revert its
behavior so that {{computeIfAbsent()}} in the case where {{async=false}} calls
"get-then-put" under the hood, so that it will (I think?) work properly with
all features?
> TestFiltering.testRandomFiltering Caffeine cache async=false
> IllegalStateException: Recursive update
> ----------------------------------------------------------------------------------------------------
>
> Key: SOLR-16707
> URL: https://issues.apache.org/jira/browse/SOLR-16707
> Project: Solr
> Issue Type: Test
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Mikhail Khludnev
> Priority: Minor
>
> see the linked mail thread.
> There are systematic test failures with async=false.
> Why can't we just turn `async=true` for this test and others?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@solr.apache.org
For additional commands, e-mail: issues-h...@solr.apache.org
