Horton

Re: Partition-wise join for join between (declaratively) partitioned tables

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>, Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>, Rafia Sabih <rafia.sabih@enterprisedb.com>, Rajkumar Raghuwanshi <rajkumar.raghuwanshi@enterprisedb.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2017-04-26T16:19:01Z
Lists: pgsql-hackers 
Robert Haas <robertmhaas@gmail.com> writes:
> I'm going to say this one more time: I really, really, really think
> you need to avoid trying to convert the partition bounds to a common
> type.  I said before that the infrastructure to do that is not present
> in our type system, and I'm pretty sure that statement is 100%
> correct.  The fact that you can find other cases where we do something
> sorta like that but in a different case with different requirements
> doesn't make that false.

It's not just a matter of lack of infrastructure: the very attempt is
flawed, because in some cases there simply isn't a supertype that can
hold all values of both types.  An easy counterexample is float8 vs
numeric: you can't convert float8 'Infinity' to numeric, but also there
are values of numeric that can't be converted to float8 without overflow
and/or loss of precision.

The whole business of precision loss makes things very touchy for almost
anything involving float and a non-float type, actually.

What I'm going to ask one more time, though, is why we are even discussing
this.  Surely the partition bounds of a partitioned table must all be of
the same type already.  If there is a case where they are not, that is
a bug we had better close off before v10 ships, not a feature that we
need to write a lot of code to accommodate.

			regards, tom lane

Commits

  1. Basic partition-wise join functionality.

  2. Assorted preparatory refactoring for partition-wise join.

  3. Teach adjust_appendrel_attrs(_multilevel) to do multiple translations.

  4. Stamp 10beta2.

  5. Eat XIDs more efficiently in recovery TAP test.

  6. Abstract logic to allow for multiple kinds of child rels.

  7. Implement SortSupport for macaddr data type

  8. Attempt to stabilize grouping sets regression test plans.

  9. Teach xlogreader to follow timeline switches

  10. Don't scan partitioned tables.

  11. Fix grammar.

  12. postgres_fdw: Push down FULL JOINs with restriction clauses.

  13. Some preliminary refactoring towards partitionwise join.

  14. contrib/amcheck needs RecentGlobalXmin to be PGDLLIMPORT'ified.

  15. Print test parameters like "foo: 123", and results like "foo = 123".