Re: [HACKERS] path toward faster partition pruning

On 9 January 2018 at 21:40, Amit Langote <Langote_Amit_f8@lab.ntt.co.jp> wrote:
> Sorry about the absence in the last few days.  I will post a new version
> addressing various review comments by the end of this week.

One more thing I discovered while troubleshooting a bug Beena reported
in the run-time partition pruning patch is that
classify_partition_bounding_keys properly does;

if (EXPR_MATCHES_PARTKEY(leftop, partattno, partexpr))
    constexpr = rightop;
else if (EXPR_MATCHES_PARTKEY(rightop, partattno, partexpr))
    constexpr = leftop;
else
    /* Clause not meant for this column. */
    continue;

for OpExpr clauses, but does not do the same for leftop for the
ScalarArrayOpExpr test.

-- 
 David Rowley                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services