Vigya Sharma created LUCENE-10476:
-------------------------------------
Summary: Modify the findMerges policy for
addIndexes(CodecReader[]) to create 1 segment per reader
Key: LUCENE-10476
URL: https://issues.apache.org/jira/browse/LUCENE-10476
Project: Lucene - Core
Issue Type: Sub-task
Reporter: Vigya Sharma
Once we have LUCENE-10216, we can trigger concurrent merges for
addIndexes(CodecReader...) api, These merges are defined in the MergePolicy and
triggered by the MergeScheduler.
We can now create a merge policy that does a 1:1 addIndexes call, where the
merger creates one segment each for every provided reader. It would create a
faster, more concurrent {{{}addIndexes(CodecReader...){}}}, at the cost of
deferring some merges to be done later in background.
Which, I believe is similar to the behavior in {{addIndexes(Directory...)}} -
all incoming segments are simply added to IW, and any merging happens in later
in background.
The change, would make both APIs behave consistently.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]