sql: grant/revoke privileges for udfs

[chengxiong-ruan]

There are 4 commits:
(1) add syntax support for grant/revoke privileges of function.
(2) implement grant/revoke privilege for functions (mostly just reading correct descriptors)
(3) fix a minor issue when hydrating a mutable function which use table implicit type (mutable table implicit type cannot be fetched)
(4) populate information_schema.role_routine_grants, and we can test privileges on that.

[cockroach-teamcity]

This change is 

[ajwerner]

I'd like to ask you to try out my suggestion in 3. Otherwise,

Reviewed 9 of 9 files at r1, 8 of 8 files at r2, 3 of 3 files at r3, 2 of 3 files at r4, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @ajwerner and @chengxiong-ruan)

---

pkg/sql/information_schema.go line 1594 at r4 (raw file):

var informationSchemaRoleRoutineGrantsTable = virtualSchemaTable{
	// TODO(chengxiong): add builtin function privileges as well.
	comment: "privileges granted on functions (incomplete; only contains privileges of user-defined functions)",

Should we file an issue to list privileges for the builtins?

---

pkg/sql/catalog/descs/hydrate.go line 268 at r3 (raw file):

	return func(ctx context.Context, id descpb.ID) (tree.TypeName, catalog.TypeDescriptor, error) {
		var typDesc catalog.TypeDescriptor
		typDesc, err := tc.GetMutableTypeVersionByID(ctx, txn, id)

What would happen if we always got an immutable type here with AvoidLeased? I have a feeling that the choice to get a Mutable type here is vestigial. I'd be curious to learn if anything breaks. I think always resolving an immutable here would be strictly better if it doesn't cause any problems.

[chengxiong-ruan]

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @ajwerner)

---

pkg/sql/information_schema.go line 1594 at r4 (raw file):

Previously, ajwerner wrote…

Should we file an issue to list privileges for the builtins?

Yes, I'll file one.

---

pkg/sql/catalog/descs/hydrate.go line 268 at r3 (raw file):

Previously, ajwerner wrote…

What would happen if we always got an immutable type here with AvoidLeased? I have a feeling that the choice to get a Mutable type here is vestigial. I'd be curious to learn if anything breaks. I think always resolving an immutable here would be strictly better if it doesn't cause any problems.

yes, I believe getting an immutable with AvoidLeased will work here since we just want the fresh metadata and not really mutate anything. I'll make the change.

[chengxiong-ruan]

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

---

pkg/sql/information_schema.go line 1594 at r4 (raw file):

Previously, chengxiong-ruan (Chengxiong Ruan) wrote…

Yes, I'll file one.

oh, I don't need to file one, I'll reuse the existing issue: #84297

---

pkg/sql/catalog/descs/hydrate.go line 268 at r3 (raw file):

Previously, chengxiong-ruan (Chengxiong Ruan) wrote…

yes, I believe getting an immutable with AvoidLeased will work here since we just want the fresh metadata and not really mutate anything. I'll make the change.

changed it to always get a fresh immutable, also changed to get immutbale db and schema.

[ajwerner]

Reviewed 1 of 3 files at r7.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @ajwerner)

[chengxiong-ruan]

TFTR!
bors r=ajwerner

[craig]

Build failed (retrying...):

• Bazel Essential CI (Cockroach)

[craig]

Build failed (retrying...):

• Bazel Essential CI (Cockroach)

[craig]

Build failed (retrying...):

• Bazel Essential CI (Cockroach)

[craig]

Build failed (retrying...):

• Bazel Essential CI (Cockroach)

[craig]

Merge conflict.

[chengxiong-ruan]

🙏
bors r+

[craig]

Build succeeded:

• Bazel Essential CI (Cockroach)