You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by "Steven Parkes (JIRA)" <ji...@apache.org> on 2007/08/06 22:11:59 UTC

[jira] Updated: (LUCENE-847) Factor merge policy out of IndexWriter

     [ https://issues.apache.org/jira/browse/LUCENE-847?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Steven Parkes updated LUCENE-847:
---------------------------------

    Attachment: LUCENE-847.patch.txt

Here's an update to the patch. I wouldn't say it's ready to be committed, but I think it's significantly closer than it was.

The concurrent and other misc. stuff have been pulled out (that part still needs work, figuring out how to get the concurrency right.)

The new patch works against trunk, which means it handles docswriter and is more compatible with merging by # of docs or merging by ram (or size, to be more accurate?)

My take on the migration path here was that we could well be going towards merging by size but need to keep merging by # docs for parallel index cases. The current patch still only does merging by # docs.

I think I commented on a couple of other things dev, but to reiterate:

There's a small change in the test results because the new merge policy simplifies the treatatement of addIndexes operations. The change is understood and shouldn't be a problem.

useCompoundFile is delegated to the merge policy so a smart merge policy could make decisions looking at the state of all segments rather than all-or-nothing. There are a couple of fixme's in IndexWriter related to this and the segments being created by the docswriter.

I'm going to look at that, plus the concurrent stuff: Ning's stuff plus by old approach (which has to change, given the new docswriter stuff).

> Factor merge policy out of IndexWriter
> --------------------------------------
>
>                 Key: LUCENE-847
>                 URL: https://issues.apache.org/jira/browse/LUCENE-847
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: Index
>            Reporter: Steven Parkes
>            Assignee: Steven Parkes
>         Attachments: concurrentMerge.patch, LUCENE-847.patch.txt, LUCENE-847.txt
>
>
> If we factor the merge policy out of IndexWriter, we can make it pluggable, making it possible for apps to choose a custom merge policy and for easier experimenting with merge policy variants.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-dev-help@lucene.apache.org