[
https://issues.apache.org/jira/browse/LENS-94?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14241011#comment-14241011
]
Amareshwari Sriramadasu commented on LENS-94:
---------------------------------------------
The issue here is because of the following code :
{noformat}
- // TODO why is only first edge?
- TableRelationship edge = path.getEdges().get(0);
{noformat}
The above code adding columns only from first edge. In the case of multiple
paths :
Ex : testcube.testdim3id->testdim3.id
testcube.dim2->testdim2.id->testdim2.testdim3id->testdim3.id.
It adds edges testcube.testdim3id->testdim3.id and
testdim2.testdim3id->testdim3.id only. Which misses dim2 column for cubes. So
we need add the all edges. Added test case for this.
And the columns of from and to columns have to be separated because candidate
resolver should discard facts/dimtables if any of from columns does not exist.
Similarly if any of to columns does not exists. If they are not separated, if
from column exist and to column does not exist, those facts/dimtables are not
getting pruned. Existing tests are fail because of this. So, no new tests
added.
The fix put up on https://reviews.apache.org/r/28849 solves the issues above.
> When there are multiple join paths then No driver accepts the query
> -------------------------------------------------------------------
>
> Key: LENS-94
> URL: https://issues.apache.org/jira/browse/LENS-94
> Project: Apache Lens
> Issue Type: Bug
> Reporter: Arshad Matin
> Assignee: Amareshwari Sriramadasu
> Attachments: LENS-94.WIP.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
