Re: [PATCH] Add pg_get_type_ddl() to retrieve the CREATE TYPE statement

Chao Li <li.evan.chao@gmail.com>

From: Chao Li <li.evan.chao@gmail.com>
To: Philip Alger <paalger0@gmail.com>
Cc: Quan Zongliang <quanzongliang@yeah.net>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2025-10-31T23:09:24Z
Lists: pgsql-hackers

> On Nov 1, 2025, at 06:29, Philip Alger <paalger0@gmail.com> wrote:
> 
> 
> <v2-0001-Add-pg_get_type_ddl-function.patch>

1
```
+	/*
+	 * Look up the type tuple to allow shell types.
+	 */
+	typeTup = LookupTypeName(NULL, typeStruct, NULL, false);
+	if (typeTup == NULL)
+		ereport(ERROR,
+				(errcode(ERRCODE_UNDEFINED_OBJECT),
+				 errmsg("type \"%s\" does not exist",
+						TypeNameToString(typeStruct))));
```

Here when you call LookupTypeName(), you give the last parameter “missing_ok” a value of “false”, so that it would “ereport” inside LookupTypeName(), so your manual check of “if (typeTup == NULL)” will never be satisfied.

2
```
+{ oid => '8414', descr => 'get CREATE statement for type',
+  proname => 'pg_get_type_ddl', prorettype => 'text', proisstrict => 't',
+  proargtypes => 'text', proargnames => '{typname}',
+  prosrc => 'pg_get_type_ddl' },
```
Here you set  proisstrict => ’t’. With strict mode, the function will not be executed if any of input arguments are NULL.

So add this test seems meaningless, because the function is not executed at all.
```
+SELECT pg_get_type_ddl(NULL);
+ pg_get_type_ddl 
+-----------------
+ 
+(1 row)
```

3. As discussed in other get_xxx_ddl() patches, does this function needs a pretty flag? I think other patches have that.

Best regards,
--
Chao Li (Evan)
HighGo Software Co., Ltd.
https://www.highgo.com/