Horton

RE: Support logical replication of DDLs

Wei Wang (Fujitsu) <wangw.fnst@fujitsu.com>

View thread
From: "wangw.fnst@fujitsu.com" <wangw.fnst@fujitsu.com>
To: Ajin Cherian <itsajin@gmail.com>, Peter Smith <smithpb2250@gmail.com>
Cc: vignesh C <vignesh21@gmail.com>, Zheng Li <zhengli10@gmail.com>, li jie <ggysxcq@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>, rajesh singarapu <rajesh.rs0541@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-03-06T10:16:59Z
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.

On Mon, Mar 6, 2023 14:34 AM Ajin Cherian <itsajin@gmail.com> wrote:
> Changes are in patch 1 and patch 2

Thanks for updating the patch set.

Here are some comments:

For v-75-0002* patch.
1. In the function deparse_AlterRelation.
+		if ((sub->address.objectId != relId &&
+			 sub->address.objectId != InvalidOid) &&
+			!(subcmd->subtype == AT_AddConstraint &&
+			  subcmd->recurse) &&
+			istable)
+			continue;

I think when we execute the command "ALTER TABLE ... ADD (index)" (subtype is
AT_AddIndexConstraint or AT_AddIndex), this command will be skipped for parsing.
I think we need to parse both types of commands here.

===
For v-75-0003* patch.
2. In the function deparse_CreateSeqStmt.
It seems that we are not deparsing the "AS data_type" clause (CREATE SEQUENCE
... AS data_type). I think this causes all data_type to be default (bigint)
after executing the parsed CreateSeq command.

~~~

3. In the function deparse_CreateTrigStmt
+		tgargs = DatumGetByteaP(fastgetattr(trigTup,
+											Anum_pg_trigger_tgargs,
+											RelationGetDescr(pg_trigger),
+											&isnull));
+		if (isnull)
+			elog(ERROR, "null tgargs for trigger \"%s\"",
+				 NameStr(trigForm->tgname));
+		argstr = (char *) VARDATA(tgargs);
+		lentgargs = VARSIZE_ANY_EXHDR(tgargs);
a.
I think it might be better to invoke the function DatumGetByteaP after checking
the flag "isnull". Because if "isnull" is true, I think an unexpected pointer
((NULL)->va_header) will be used when invoking macro VARATT_IS_4B_U.

b.
Since commit 3a0d473 recommends using macro DatumGetByteaPP instead of
DatumGetByteaP, and the function pg_get_triggerdef_worker also uses macro
DatumGetByteaPP, I think it might be better to use DatumGetByteaPP here.

~~~

4. In the function deparse_CreateTrigStmt
+	append_object_object(ret, "EXECUTE PROCEDURE %{function}s", tmp_obj);

Since the use of the keyword "PROCEDURE" is historical, I think it might be
better to use "FUNCTION".

===
For v-75-0004* patch.
5. File src/test/modules/test_ddl_deparse_regress/README.md
+1. Test that the generated JSON blob is expected using SQL tests.
+2. Test that the re-formed DDL command is expected using SQL tests.
+3. Test	that the re-formed DDL command has the same effect as the original command
+   by comparing	the results of pg_dump,	using the SQL tests in 1 and 2.
+4. Test	that new DDL syntax is handled by the DDL deparser by capturing and deparing
+   DDL commands	ran by pg_regress.

Inconsistent spacing:
\t -> blank space

Regards,
Wang wei