[
https://issues.apache.org/jira/browse/LUCENE-3292?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13062809#comment-13062809
]
Michael McCandless commented on LUCENE-3292:
--------------------------------------------
SegmentCacheKey.equals needs { } around the single-statement ifs.
Make sure you remove the TODO in ReaderPool.get since this issue is fixing it.
{quote}
I am still not sure on 2 things:
In SegmentReader#get the assert statement.
{quote}
Hmm... why are we adding this assert again?
bq. In IW.ReaderPool.SegmentCacheKey the equals() method - Does it need to call
.equals on the SegmetnInfo and context? Since we are just including
IOContext.Context only won't this be wrong ?
Yes, it should && the .equals of its two member. It's not wrong -- we only
must ensure that the SegmentReader for merging is shared during merging, and
for searching during searching; we don't want t open a new SegmentReader for
every new IOContext.
Actually: why do we have IOContext.equals/hashCode...? Who relies on these
methods today?
ReaderPool.commit should pass IOContext.READ.
I think we should add a ReaderPool.getIfExists that does not take an IOContext
and tries both READ and MERGE and returns the SegmentReader if it's found;
likewise for release.
> IOContext should be part of the SegmentReader cache key
> --------------------------------------------------------
>
> Key: LUCENE-3292
> URL: https://issues.apache.org/jira/browse/LUCENE-3292
> Project: Lucene - Java
> Issue Type: Task
> Components: core/index
> Affects Versions: 4.0
> Reporter: Simon Willnauer
> Assignee: Varun Thacker
> Priority: Minor
> Fix For: 4.0
>
> Attachments: LUCENE-3292.patch, LUCENE-3292.patch
>
>
> Once IOContext (LUCENE-2793) is landed the IOContext should be part of the
> key used to cache that reader in the pool
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
