plpgsql: add support for INTO STRICT option

[DrewKimball]

logictest: move plpgsql SELECT/RETURNING INTO tests to new file

This commit is test-only refactor that moves the tests for SELECT
and RETURNING INTO PL/pgSQL syntax to their own file.

Epic: None

Release note: None

plpgsql: add support for INTO STRICT syntax

This commit adds support for SELECT INTO STRICT and
RETURNING INTO STRICT for SQL statements within a PL/pgSQL routine.
This option causes the routine to error when the SQL statement doesn't
return exactly one row. This is handled via a ScalarGroupBy to compute
the row count, followed by a call to crdb_internal.plpgsql_raise if
necessary to throw the error.

Fixes #115297

Release note (sql change): It is now possible to use the STRICT option
with SELECT ... INTO and RETURNING INTO in order to enforce that the
SQL statement returns exactly one row.

sql: add setting to make PL/pgSQL INTO strict by default

This commit adds a new session setting, plpgsql_use_strict_into, that
causes PL/pgSQL SELECT ... INTO and RETURNING INTO syntax to behave
as if the STRICT option was specified. This causes the SQL statement
to result in an error if it doesn't return exactly one row. This is
useful for applications migrating from oracle, since it has this "strict"
behavior by default.

Informs #115297

Release note (sql change): Added a setting plpgsql_use_strict_into,
which causes PL/pgSQL SELECT ... INTO and RETURNING INTO to require
exactly one row from the SQL statement, similar to oracle behavior.

[blathers-crl]

Your pull request contains more than 1000 changes. It is strongly encouraged to split big PRs into smaller chunks.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[cockroach-teamcity]

This change is 

[yuzefovich]

Nice!

Reviewed 19 of 19 files at r1, 2 of 2 files at r2, 10 of 10 files at r3, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @DrewKimball and @mgartner)

---

-- commits line 34 at r3:
For my own edification, how do you opt out of STRICT default behavior in Oracle? Via a similar setting that disables "strict by default"?

---

pkg/sql/vars.go line 3243 at r3 (raw file):

			return nil
		},
		GlobalDefault: globalFalse,

nit: should we introduce a cluster setting too?

---

pkg/sql/opt/optbuilder/plpgsql.go line 1014 at r2 (raw file):

}

// buildPLpgSQLRaise builds a Project expression which calls the implements the

nit: "which calls the implements ..."

---

pkg/ccl/logictestccl/testdata/logic_test/plpgsql_into line 398 at r2 (raw file):

query II rowsort
SELECT * FROM xy;

nit: might be a bit less verbose to use SELECT count(*) FROM xy; or SELECT count(*) == 8 FROM xy; since we only care whether another row was inserted or not.

[DrewKimball]

TFYR!

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @mgartner and @yuzefovich)

---

-- commits line 34 at r3:

Previously, yuzefovich (Yahor Yuzefovich) wrote…

For my own edification, how do you opt out of STRICT default behavior in Oracle? Via a similar setting that disables "strict by default"?

The docs claim this behavior is "by default", but I haven't been able to find a way to do that. See the "Usage Notes" section in the docs. I guess maybe you could use LIMIT 1 plus an exception handler that catches NO_DATA_FOUND and sets variables to NULL?

---

pkg/sql/vars.go line 3243 at r3 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: should we introduce a cluster setting too?

I think it's probably fine to leave as-is, since the default can be set for a user (right?). I'm open to convincing otherwise, though.

---

pkg/sql/opt/optbuilder/plpgsql.go line 1014 at r2 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: "which calls the implements ..."

Done.

---

pkg/ccl/logictestccl/testdata/logic_test/plpgsql_into line 398 at r2 (raw file):

Previously, yuzefovich (Yahor Yuzefovich) wrote…

nit: might be a bit less verbose to use SELECT count(*) FROM xy; or SELECT count(*) == 8 FROM xy; since we only care whether another row was inserted or not.

Good idea, done.

[yuzefovich]

Reviewed 10 of 10 files at r4, 10 of 10 files at r6, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @DrewKimball and @mgartner)

---

pkg/sql/vars.go line 3243 at r3 (raw file):

Previously, DrewKimball (Drew Kimball) wrote…

I think it's probably fine to leave as-is, since the default can be set for a user (right?). I'm open to convincing otherwise, though.

Yeah, sgtm.

[DrewKimball]

bors r+

[craig]

Build failed:

• Bazel Essential CI (Cockroach)

[DrewKimball]

bors r+

[craig]

Build succeeded:

• Bazel Essential CI (Cockroach)
• macos_amd64_build
• macos_arm64_build
• windows_build

[mgartner]

Nicely done!!!

Reviewed 1 of 19 files at r1, 27 of 27 files at r7, 2 of 2 files at r8, 10 of 10 files at r9, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (and 1 stale)