Horton

Re: SQL:2011 application time

Paul A Jungwirth <pj@illuminatedcomputing.com>

View thread
From: Paul Jungwirth <pj@illuminatedcomputing.com>
To: Peter Eisentraut <peter@eisentraut.org>
Cc: jian he <jian.universality@gmail.com>, Isaac Morland <isaac.morland@gmail.com>, Robert Haas <robertmhaas@gmail.com>, Jeff Davis <pgsql@j-davis.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2024-10-21T21:46:14Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Rename gist stratnum support function

  2. Remove support for temporal RESTRICT foreign keys

  3. Cache NO ACTION foreign keys separately from RESTRICT foreign keys

  4. Fix NO ACTION temporal foreign keys when the referenced endpoints change

  5. Improve whitespace in without_overlaps test

  6. Tests for logical replication with temporal keys

  7. Support for GiST in get_equal_strategy_number()

  8. Make the conditions in IsIndexUsableForReplicaIdentityFull() more explicit

  9. Replace get_equal_strategy_number_for_am() by get_equal_strategy_number()

  10. Improve internal logical replication error for missing equality strategy

  11. Simplify IsIndexUsableForReplicaIdentityFull()

  12. Fix ALTER TABLE / REPLICA IDENTITY for temporal tables

  13. doc: Update pg_constraint.conexclop docs for WITHOUT OVERLAPS

  14. doc: Add PERIOD to ALTER TABLE reference docs

  15. doc: Add WITHOUT OVERLAPS to ALTER TABLE reference docs

  16. Add temporal FOREIGN KEY contraints

  17. Add temporal PRIMARY KEY and UNIQUE constraints

  18. Add stratnum GiST support function

  19. Avoid crashing when a JIT-inlined backend function throws an error.

  20. Revert temporal primary keys and foreign keys

  21. Fix ON CONFLICT DO NOTHING/UPDATE for temporal indexes

  22. Add test for REPLICA IDENTITY with a temporal key

  23. Use half-open interval notation in without_overlaps tests

  24. Use daterange and YMD in without_overlaps tests instead of tsrange.

  25. Rename pg_constraint.conwithoutoverlaps to conperiod

  26. Fix comment on gist_stratnum_btree

  27. Add missing TAP test name

  28. Improve error handling of HMAC computations

  29. Rename functions to avoid future conflicts

Attachments

Hello,

Here is a new set of patches, including new patches to (1) fix logical replication with WITHOUT 
OVERLAPS indexes and (2) address some documentation lapses pointed out in jian he's feedback. Since 
all that is against the already-commited PK/UNIQUE/FK work, I've kept them separate here from the 
FOR PORTION OF etc patches. I've also added the logical replication problem to the v18 Open Items 
wiki page.

Details below:

On 8/15/24 19:12, jian he wrote:
 > in create_table.sgml, I saw the WITHOUT OVERLAPS change is mainly in
 > table_constraint.
 > but we didn't touch alter_table.sgml.
 > Do we also need to change alter_table.sgml correspondingly?

Good catch! Not just WITHOUT OVERLAPS but FOREIGN KEY needed to be updated too. Patched here.

 > + if (constraint->without_overlaps)
 > + {
 > + /*
 > + * This enforces that there is at least one equality column
 > + * besides the WITHOUT OVERLAPS columns.  This is per SQL
 > + * standard.  XXX Do we need this?
 > + */
 > + if (list_length(constraint->keys) < 2)
 > + ereport(ERROR,
 > + errcode(ERRCODE_SYNTAX_ERROR),
 > + errmsg("constraint using WITHOUT OVERLAPS needs at least two columns"));
 > +
 > + /* WITHOUT OVERLAPS requires a GiST index */
 > + index->accessMethod = "gist";
 > + }
 > if Constraint->conname is not NULL, we can
 > + errmsg("constraint \"%s\" using WITHOUT OVERLAPS needs at least two
 > columns"));
 >
 > "XXX Do we need this?"
 > I think currently we need this, otherwise the following create_table
 > synopsis will not be correct.
 > UNIQUE [ NULLS [ NOT ] DISTINCT ] ( column_name [, ... ] [,
 > column_name WITHOUT OVERLAPS ] )
 > PRIMARY KEY ( column_name [, ... ] [, column_name WITHOUT OVERLAPS ] )

Having zero scalar columns is not allowed according to SQL:2011, but supporting it seems like a cool 
thing to do eventually. It doesn't have to be in this first set of patches. But if we support it 
someday, I agree we must update these docs to match.

 > we add a column in catalog-pg-constraint.
 > do we need change column conexclop,
 > "If an exclusion constraint, list of the per-column exclusion operators"
 > but currently, primary key, unique constraint both have valid conexclop.

I updated the docs to include the WITHOUT OVERLAPS case.

 > +static void
 > +ExecWithoutOverlapsNotEmpty(Relation rel, NameData attname, Datum
 > attval, char typtype, Oid atttypid)
 > ...
 > I think in the default branch, you need at least set the isempty
 > value, otherwise maybe there will be a compiler warning
 > because later your use isempty, but via default branch is value undefined?

This isn't getting a warning on the build farm, so I think compilers are smart enough to see that 
elog exits.

 > + /*
 > + * If this is a WITHOUT OVERLAPS constraint,
 > + * we must also forbid empty ranges/multiranges.
 > + * This must happen before we look for NULLs below,
 > + * or a UNIQUE constraint could insert an empty
 > + * range along with a NULL scalar part.
 > + */
 > + if (indexInfo->ii_WithoutOverlaps)
 > + {
 > +             ExecWithoutOverlapsNotEmpty(heap, att->attname,
 > + }
 > previously we found out that if this happens later, then it won't work.
 > but this comment didn't explain why this must have happened earlier.
 > I didn't dig deep enough to find out why.
 > but explaining it would be very helpful.

I don't follow. Explaining why it can't happen later is the same as explaining why it must happen 
sooner, right? But you say "must *have* happened earlier"---it's not that it must have happened 
already, but that we must do it now (before bailing when we look for NULLs).

 > I think some tests are duplicated, so I did the refactoring.

I kept some of this patch: querying tableoid::regclass from the partition root is nicer than 
querying the leaves separately.

The changes removing DELETEs & INSERTs made the tests less local, since they would depend on inserts 
made further up in the file. I'd rather keep the tests fairly isolated. I don't think this is the 
same thing as duplication.

Rebased to 68ad9816c1.

Yours,

-- 
Paul              ~{:-)
pj@illuminatedcomputing.com