jackrabbit-oak-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vikas Saurabh (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (OAK-5927) Load excerpt lazily
Date Fri, 17 Mar 2017 08:29:41 GMT

    [ https://issues.apache.org/jira/browse/OAK-5927?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15929577#comment-15929577

Vikas Saurabh commented on OAK-5927:

Actually, it seems to me that in both those cases, the underlying issue is that the cursor
is doing more work as it is/can not take the context of execution into consideration. Like
why should getSize run with the same columns - it only needs to take care of constraints.
For pagination: seeking to 4th page with 100 results per page should, Imo, be taken as treat
401st row as first row ie fetch 401-450 then 451-550 etc.

Btw, I'm not saying we shouldn't do this one - just that I think this is pointing to some
other improvements that are missing.

> Load excerpt lazily
> -------------------
>                 Key: OAK-5927
>                 URL: https://issues.apache.org/jira/browse/OAK-5927
>             Project: Jackrabbit Oak
>          Issue Type: Improvement
>          Components: lucene
>            Reporter: Chetan Mehrotra
>              Labels: performance
>             Fix For: 1.8
> Currently LucenePropertyIndex loads the excerpt eagerly in batch as part of loadDocs
call. The load docs batch size doubles starting from 50 (max 100k) as more data is read. 
> We should look into ways to make the excerpt loaded lazily as and when caller ask for
> Note that currently the excerpt are only loaded when query request for excerpt i.e. there
is a not null property restriction for {{rep:excerpt}}. 

This message was sent by Atlassian JIRA

View raw message