For these types of changes, I think we need to have a broader discussion before merging commits. We've spent a substantial amount of time in the last six months trying to fix a large number of bugs that were in decorrelation. I'm not surprised that more have been found as it is very complicated code. While this particular change may address this bug, I think it introduces a number of other issues. I'm all for quick fixes and minimal process around those. However, if someone is working with code that someone else has put a substantial amount of time into, it would be great to make sure everyone is on the same page before committing.
Could you please share a more in-depth analysis to why you think removing this was the right fix? E.g. what the symptoms were and where you saw decorrelation failing with this enabled. We'd be happy to work through a more holistic fix. On Tue, Jul 8, 2014 at 6:46 PM, Julian Hyde (JIRA) <[email protected]> wrote: > > [ > https://issues.apache.org/jira/browse/OPTIQ-313?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel > ] > > Julian Hyde resolved OPTIQ-313. > ------------------------------- > > Resolution: Fixed > > Fixed in > http://git-wip-us.apache.org/repos/asf/incubator-optiq/commit/58ddf6bf. > > > Query decorrelation fails > > ------------------------- > > > > Key: OPTIQ-313 > > URL: https://issues.apache.org/jira/browse/OPTIQ-313 > > Project: Optiq > > Issue Type: Bug > > Reporter: Julian Hyde > > Assignee: Julian Hyde > > > > TPC-H query 2 fails during preparation because even after decorrelation > there is a CorrelatorRel present. > > > > -- > This message was sent by Atlassian JIRA > (v6.2#6252) >
