Re: Support logical replication of DDLs

Zheng Li <zhengli10@gmail.com>

From: Zheng Li <zhengli10@gmail.com>
To: li jie <ggysxcq@gmail.com>
Cc: vignesh C <vignesh21@gmail.com>, Ajin Cherian <itsajin@gmail.com>, rajesh singarapu <rajesh.rs0541@gmail.com>, Peter Smith <smithpb2250@gmail.com>, Dilip Kumar <dilipbalaut@gmail.com>, Alvaro Herrera <alvherre@alvh.no-ip.org>, "houzj.fnst@fujitsu.com" <houzj.fnst@fujitsu.com>, Amit Kapila <amit.kapila16@gmail.com>, Masahiko Sawada <sawada.mshk@gmail.com>, Japin Li <japinli@hotmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2022-12-02T02:46:25Z
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. Add a run_as_owner option to subscriptions.

  2. Refactor pgoutput_change().

  3. Print the correct aliases for DML target tables in ruleutils.

  4. Fix object identity string for transforms

  5. Add grantable MAINTAIN privilege and pg_maintain role.

  6. Get rid of recursion-marker values in enum AlterTableType

  7. Release cache tuple when no longer needed

  8. Empty search_path in logical replication apply worker and walsender.

  9. Refactor format_type APIs to be more modular

  10. Use wrappers of PG_DETOAST_DATUM_PACKED() more.

Attachments

> I applied patch 0005.
>
> I think this modification is a bit overdone.
> This design skips all subcommands, which results in many ddl
> replication failures.
> For example:
> ```
> CREATE TABLE datatype_table (id SERIAL);
> ```
> deparsed ddl is:
> CREATE  TABLE  public.datatype_table (id pg_catalog.int4 STORAGE plain
> NOT NULL DEFAULT
> pg_catalog.nextval('public.datatype_table_id_seq'::pg_catalog.regclass))
> CREATE SEQUENCE subcommand will be skipped.
>
> OR:
> ```
> CREATE SCHEMA element_test
> CREATE TABLE foo (id int)
> CREATE VIEW bar AS SELECT * FROM foo;
> ```
> deparsed ddl is:
> CREATE SCHEMA element_test.
>
> Its subcommands will be skipped.
> There may be other cases.
>
> For the initial CREATE LIKE statement, It is special,
> It derives the subcommand of alter table column.
> Just skipping them may be enough.
> Instead of skipping subcommands of all statements.
> After all, our design is to obtain the actual ddl information from the
> catalog instead of parsing raw parsetree.
> This is why we cannot skip all subcommands.

Agreed, event triggers capture commands on the subcommand level so
we can't skip subcommands at large without modifying how to
deparse top level commands such as CREATE TABLE/INDEX/SCHEMA.
Also for special statements like CREATE TABLE datatype_table (id SERIAL);
the CREATE SEQUENCE subcommand actually has to execute before
the top level command CREATE TABLE.

Attached please find a new solution that skips the deparsing of ALTER TABLE
subcommands generated for TableLikeClause. The patch v42-0005 added a new
boolean field table_like to AlterTableStmt in order to identify an ALTER TABLE
subcommand generated internally for the TableLikeClause.

Regards,
Zheng