[
https://issues.apache.org/jira/browse/JENA-619?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13864055#comment-13864055
]
Andy Seaborne commented on JENA-619:
------------------------------------
One possibility is to split filter placement into two phases - all algebra
operations except BGPs, then a separate pass to place within, and breaking up,
BGPs. This allows controlling the reordering of triple/quads patterns by the
storage layer, which may have more detailed information, and the filter
placement process.
See
[users@
enquiry|http://mail-archives.apache.org/mod_mbox/jena-users/201401.mbox/%3CEMEW3|890578526c59c0dea4fc317a030a0502q05CZZ07sza1g10|ecs.soton.ac.uk|62403AD2-AD43-468B-9269-87A106AA1DD5%40ecs.soton.ac.uk%3E]
and the
[reply|http://mail-archives.apache.org/mod_mbox/jena-users/201401.mbox/%3C52CBCC3C.9050404%40apache.org%3E].
> Optimization of filter placement and BGP reordering.
> ----------------------------------------------------
>
> Key: JENA-619
> URL: https://issues.apache.org/jira/browse/JENA-619
> Project: Apache Jena
> Issue Type: Improvement
> Reporter: Andy Seaborne
>
> As a follow-on from JENA-595, now consider the interaction of filter
> placement and BGP reordering.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
