On Sun, Feb 18, 2024 at 7:29 AM Tomas Vondra
<tomas.vondra@enterprisedb.com> wrote:
>
>
> Also, now that I looked at the v2 patch again, I see it only really
> tweaked the pg_proc.dat entry, but the code still does PG_GETARG_OID (so
> the "any" bit is not really correct).
>
PG_GETARG_OID part indeed is wrong. so I change to following:
+Datum
+pg_basetype(PG_FUNCTION_ARGS)
+{
+ Oid oid;
+
+ oid = get_fn_expr_argtype(fcinfo->flinfo, 0);
+ if (!SearchSysCacheExists1(TYPEOID, ObjectIdGetDatum(oid)))
+ PG_RETURN_NULL();
+
+ PG_RETURN_OID(getBaseType(oid));
+}
I still name the function as pg_basetype, feel free to change it.
+ <row>
+ <entry role="func_table_entry"><para role="func_signature">
+ <indexterm>
+ <primary>pg_basetype</primary>
+ </indexterm>
+ <function>pg_basetype</function> ( <type>"any"</type> )
+ <returnvalue>regtype</returnvalue>
+ </para>
+ <para>
+ Returns the OID of the base type of a domain or if the
argument is a basetype it returns the same type.
+ If there's a chain of domain dependencies, it will recurse
until finding the base type.
+ </para>
compared with pg_typeof's explanation, I feel like pg_basetype's
explanation doesn't seem accurate.
However, I don't know how to rephrase it.