[
https://issues.apache.org/jira/browse/LUCENE-10129?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17429252#comment-17429252
]
Uwe Schindler edited comment on LUCENE-10129 at 10/15/21, 12:04 PM:
--------------------------------------------------------------------
bq. I am wondering though if TestRamUsageEstimator is missing an import static
org.apache.lucene.util.RamUsageEstimator.sizeOf;, so that in lines like
assertEquals(sizeOf(array), sizeOf((Object) array)); the first sizeOf() calls
RamUsageEstimator.sizeOf, and the second calls RamUsageTester.sizeOf. Apologies
if I misunderstood the purpose of the test.
I was also stumbling on this. Maybe we should remove the static import and be
explicit and compare all three versions?
was (Author: thetaphi):
bq. I am wondering though if TestRamUsageEstimator is missing an import static
org.apache.lucene.util.RamUsageEstimator.sizeOf;, so that in lines like
assertEquals(sizeOf(array), sizeOf((Object) array)); the first sizeOf() calls
RamUsageEstimator.sizeOf, and the second calls RamUsageTester.sizeOf.
Apologies if I misunderstood the purpose of the test.
I was also stumbling on this. Maybe we should remove the static import and be
explicit and compare all three versions?
> Add RamUsageEstimator shallowSizeOf(long[]) overload that just calls
> sizeOf(long[])?
> ------------------------------------------------------------------------------------
>
> Key: LUCENE-10129
> URL: https://issues.apache.org/jira/browse/LUCENE-10129
> Project: Lucene - Core
> Issue Type: Task
> Reporter: Robert Muir
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> See LUCENE-10128 for an example. The problem is there is only a
> {{sizeOf(long[])}}, so if the programmer uses {{shallowSizeOf}} instead of
> {{sizeOf}} then it falls back to {{shallowSizeOf(Object)}} which does a bunch
> of reflection.
> This is pretty crazy because it can create performance traps. Should we just
> add a {{shallowSizeOf(long[])}} that calls {{sizeOf(long[])}}, so that things
> are fast? (same for other primitive arrays). It would solve the problem
> easily I think.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
