trafodion-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eric Owhadi <>
Subject question for predicate pushdown feature.
Date Tue, 15 Dec 2015 22:15:33 GMT
I was running regression test on the new predicate pushdown feature, and
stumbled on a bug that I now understand:

When deciding that a predicate can be pushed down or not, we are checking
if it is in the form <column><op><value>, along with various other limiting
conditions (this is also true for the current existing pushdown feature
that also have this problem).

However, I discovered that if we have something that would trigger a scan,
and that the predicate on the scan refers to the column on a different
table than the table being scanned, we incorrectly push down as if the
column involved belong to the table being scanned.

I wanted to implement a fix, by checking if the table associated with the
column, is the same as the column in the scan, but I am facing an issue
where NATable associated with NAColumn is optional, so can be NULL.

Anyone know if I should never face the NULL situation in the case of
predicate on a SCAN object?

Or is there another way to solve this problem?


  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message