Horton

Re: Declarative partitioning - another take

Robert Haas <robertmhaas@gmail.com>

View thread View on postgr.es
From: Robert Haas <robertmhaas@gmail.com>
To: Francisco Olarte <folarte@peoplecall.com>
Cc: Corey Huinker <corey.huinker@gmail.com>, Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>, Amit Langote <amitlangote09@gmail.com>, Rajkumar Raghuwanshi <rajkumar.raghuwanshi@enterprisedb.com>, Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>, Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2016-11-01T18:09:26Z
Lists: pgsql-hackers 
On Tue, Nov 1, 2016 at 2:05 PM, Francisco Olarte <folarte@peoplecall.com> wrote:
>> /me raises hand.  We have tables with a taxonomy in them where the even data
>> splits don't fall on single letter boundaries, and often the single string
>> values have more rows than entire letters. In those situations, being able
>> to express ['XYZ','XYZ'] is important.  ['XYZ,'XZ') would let 'XYZ1' bleed
>> into the partition and ['XYZ','XYZ1') lets in other values, and so I go
>> chasing down the non-discrete set rabbit hole.
>
> You just do the classical ( I've had to do it ) closed end || minimum
> char ( "XYZ","XYZ\0" in this case ). It is not that difficult as
> strings have a global order, the next string to any one is always that
> plus the \0, or whatever your minimum is.

In defense of Corey's position, that's not so easy.  First, \0 doesn't
work; our strings can't include null bytes.  Second, the minimum legal
character depends on the collation in use.  It's not so easy to figure
out what the "next" string is, even though there necessarily must be
one.

> The problem is with anything similar to a real number, but then there
> I've always opted for half-open interval, as they can cover the line
> without overlapping, unlike closed ones.
>
> Anyway, as long as anyone makes sure HALF-OPEN intervals are allowed,
> I'm fine ( I do not remember the name, but once had to work with a
> system that only allowed closed or open and it was a real PITA.

I think we're all in agreement that half-open intervals should not
only be allowed, but the default.  The question is whether it's a good
idea to also allow other possibilities.

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Commits

  1. Fix typo.

  2. Document trigger-firing behavior for inheritance/partitioning.

  3. Fire per-statement triggers on partitioned tables.

  4. Set ecxt_scantuple correctly for tuple routing.

  5. Fix interaction of partitioned tables with BulkInsertState.

  6. Avoid core dump for empty prepared statement in an aborted transaction.

  7. Fix some problems in check_new_partition_bound().

  8. Remove unnecessary arguments from partitioning functions.

  9. Fix reporting of constraint violations for table partitioning.

  10. Fix tuple routing in cases where tuple descriptors don't match.

  11. Invalid parent's relcache after CREATE TABLE .. PARTITION OF.

  12. Doc: improve documentation about inheritance.