Adjust collation determination rules as per discussion.

Remove crude hack that tried to propagate collation through a
function-returning-record, ie, from the function's arguments to individual
fields selected from its result record.  That is just plain inconsistent,
because the function result is composite and cannot have a collation;
and there's no hope of making this kind of action-at-a-distance work
consistently.  Adjust regression test cases that expected this to happen.

Meanwhile, the behavior of casting to a domain with a declared collation
stays the same as it was, since that seemed to be the consensus.

src/backend/parser/parse_collate.c

@@ -289,10 +289,11 @@ assign_collations_walker(Node *node, assign_collations_context *context)
 		case T_FieldSelect:
 			{
 				/*
-				 * FieldSelect is a special case because the field may have
-				 * a non-default collation, in which case we should use that.
-				 * The field's collation was already looked up and saved
-				 * in the node.
+				 * For FieldSelect, the result has the field's declared
+				 * collation, independently of what happened in the arguments.
+				 * (The immediate argument must be composite and thus not
+				 * collatable, anyhow.)  The field's collation was already
+				 * looked up and saved in the node.
 				 */
 				FieldSelect *expr = (FieldSelect *) node;
 
@@ -304,24 +305,6 @@ assign_collations_walker(Node *node, assign_collations_context *context)
 				if (OidIsValid(expr->resultcollid))
 				{
 					/* Node's result type is collatable. */
-					if (expr->resultcollid == DEFAULT_COLLATION_OID)
-					{
-						/*
-						 * The immediate input node necessarily yields a
-						 * composite type, so it will have no exposed
-						 * collation.  However, if we are selecting a field
-						 * from a function returning composite, see if we
-						 * can bubble up a collation from the function's
-						 * input.  XXX this is a bit of a hack, rethink ...
-						 */
-						if (IsA(expr->arg, FuncExpr))
-						{
-							FuncExpr *fexpr = (FuncExpr *) expr->arg;
-
-							if (OidIsValid(fexpr->inputcollid))
-								expr->resultcollid = fexpr->inputcollid;
-						}
-					}
 					/* Pass up field's collation as an implicit choice. */
 					collation = expr->resultcollid;
 					strength = COLLATE_IMPLICIT;

src/backend/parser/parse_func.c

@@ -1384,7 +1384,7 @@ ParseComplexProjection(ParseState *pstate, char *funcname, Node *first_arg,
 			fselect->fieldnum = i + 1;
 			fselect->resulttype = att->atttypid;
 			fselect->resulttypmod = att->atttypmod;
-			/* resultcollid may get overridden by parse_collate.c */
+			/* save attribute's collation for parse_collate.c */
 			fselect->resultcollid = att->attcollation;
 			return (Node *) fselect;
 		}

src/backend/parser/parse_target.c

@@ -1290,7 +1290,7 @@ ExpandRowReference(ParseState *pstate, Node *expr,
 		fselect->fieldnum = i + 1;
 		fselect->resulttype = att->atttypid;
 		fselect->resulttypmod = att->atttypmod;
-		/* resultcollid may get overridden by parse_collate.c */
+		/* save attribute's collation for parse_collate.c */
 		fselect->resultcollid = att->attcollation;
 
 		if (targetlist)

src/test/regress/expected/collate.linux.utf8.out

@@ -771,33 +771,33 @@ SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM collate_test3)) ORDER
  äbc
 (4 rows)
 
-CREATE FUNCTION dup (f1 anyelement, f2 out anyelement, f3 out anyarray)
-    AS 'select $1, array[$1,$1]' LANGUAGE sql;
-SELECT a, (dup(b)).* FROM collate_test1 ORDER BY 2;
- a | f2  |    f3     
----+-----+-----------
- 1 | abc | {abc,abc}
- 4 | ABC | {ABC,ABC}
- 2 | äbc | {äbc,äbc}
- 3 | bbc | {bbc,bbc}
-(4 rows)
-
-SELECT a, (dup(b)).* FROM collate_test2 ORDER BY 2;
- a | f2  |    f3     
----+-----+-----------
- 1 | abc | {abc,abc}
- 4 | ABC | {ABC,ABC}
- 3 | bbc | {bbc,bbc}
- 2 | äbc | {äbc,äbc}
-(4 rows)
-
-SELECT a, (dup(b)).* FROM collate_test3 ORDER BY 2;
- a | f2  |    f3     
----+-----+-----------
- 4 | ABC | {ABC,ABC}
- 1 | abc | {abc,abc}
- 3 | bbc | {bbc,bbc}
- 2 | äbc | {äbc,äbc}
+CREATE FUNCTION dup (anyelement) RETURNS anyelement
+    AS 'select $1' LANGUAGE sql;
+SELECT a, dup(b) FROM collate_test1 ORDER BY 2;
+ a | dup 
+---+-----
+ 1 | abc
+ 4 | ABC
+ 2 | äbc
+ 3 | bbc
+(4 rows)
+
+SELECT a, dup(b) FROM collate_test2 ORDER BY 2;
+ a | dup 
+---+-----
+ 1 | abc
+ 4 | ABC
+ 3 | bbc
+ 2 | äbc
+(4 rows)
+
+SELECT a, dup(b) FROM collate_test3 ORDER BY 2;
+ a | dup 
+---+-----
+ 4 | ABC
+ 1 | abc
+ 3 | bbc
+ 2 | äbc
 (4 rows)
 
 -- indexes

src/test/regress/expected/collate.out

@@ -502,24 +502,24 @@ SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM collate_test2)) ORDER
  bbc
 (4 rows)
 
-CREATE FUNCTION dup (f1 anyelement, f2 out anyelement, f3 out anyarray)
-    AS 'select $1, array[$1,$1]' LANGUAGE sql;
-SELECT a, (dup(b)).* FROM collate_test1 ORDER BY 2;
- a | f2  |    f3     
----+-----+-----------
- 4 | ABD | {ABD,ABD}
- 2 | Abc | {Abc,Abc}
- 1 | abc | {abc,abc}
- 3 | bbc | {bbc,bbc}
+CREATE FUNCTION dup (anyelement) RETURNS anyelement
+    AS 'select $1' LANGUAGE sql;
+SELECT a, dup(b) FROM collate_test1 ORDER BY 2;
+ a | dup 
+---+-----
+ 4 | ABD
+ 2 | Abc
+ 1 | abc
+ 3 | bbc
 (4 rows)
 
-SELECT a, (dup(b)).* FROM collate_test2 ORDER BY 2;
- a | f2  |    f3     
----+-----+-----------
- 4 | ABD | {ABD,ABD}
- 2 | Abc | {Abc,Abc}
- 1 | abc | {abc,abc}
- 3 | bbc | {bbc,bbc}
+SELECT a, dup(b) FROM collate_test2 ORDER BY 2;
+ a | dup 
+---+-----
+ 4 | ABD
+ 2 | Abc
+ 1 | abc
+ 3 | bbc
 (4 rows)
 
 -- indexes

src/test/regress/sql/collate.linux.utf8.sql

@@ -242,12 +242,12 @@ SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM collate_test1)) ORDER
 SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM collate_test2)) ORDER BY 1;
 SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM collate_test3)) ORDER BY 1;
 
-CREATE FUNCTION dup (f1 anyelement, f2 out anyelement, f3 out anyarray)
-    AS 'select $1, array[$1,$1]' LANGUAGE sql;
+CREATE FUNCTION dup (anyelement) RETURNS anyelement
+    AS 'select $1' LANGUAGE sql;
 
-SELECT a, (dup(b)).* FROM collate_test1 ORDER BY 2;
-SELECT a, (dup(b)).* FROM collate_test2 ORDER BY 2;
-SELECT a, (dup(b)).* FROM collate_test3 ORDER BY 2;
+SELECT a, dup(b) FROM collate_test1 ORDER BY 2;
+SELECT a, dup(b) FROM collate_test2 ORDER BY 2;
+SELECT a, dup(b) FROM collate_test3 ORDER BY 2;
 
 
 -- indexes

src/test/regress/sql/collate.sql

@@ -168,11 +168,11 @@ SELECT a, CAST(b AS varchar) FROM collate_test2 ORDER BY 2;
 SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM collate_test1)) ORDER BY 1;
 SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM collate_test2)) ORDER BY 1;
 
-CREATE FUNCTION dup (f1 anyelement, f2 out anyelement, f3 out anyarray)
-    AS 'select $1, array[$1,$1]' LANGUAGE sql;
+CREATE FUNCTION dup (anyelement) RETURNS anyelement
+    AS 'select $1' LANGUAGE sql;
 
-SELECT a, (dup(b)).* FROM collate_test1 ORDER BY 2;
-SELECT a, (dup(b)).* FROM collate_test2 ORDER BY 2;
+SELECT a, dup(b) FROM collate_test1 ORDER BY 2;
+SELECT a, dup(b) FROM collate_test2 ORDER BY 2;
 
 
 -- indexes