*
*
* IDENTIFICATION
- * $PostgreSQL: pgsql/src/backend/parser/parse_coerce.c,v 2.113 2003/12/17 19:49:39 tgl Exp $
+ * $PostgreSQL: pgsql/src/backend/parser/parse_coerce.c,v 2.114 2004/03/15 01:13:40 tgl Exp $
*
*-------------------------------------------------------------------------
*/
if (can_coerce_type(1, &exprtype, &targettype, ccontext))
expr = coerce_type(pstate, expr, exprtype, targettype,
ccontext, cformat);
- else if (ccontext >= COERCION_ASSIGNMENT)
- {
- /*
- * String hacks to get transparent conversions for char and
- * varchar: if a coercion to text is available, use it for forced
- * coercions to char(n) or varchar(n) or domains thereof.
- *
- * This is pretty grotty, but seems easier to maintain than providing
- * entries in pg_cast that parallel all the ones for text.
- */
- Oid targetbasetype = getBaseType(targettype);
-
- if (targetbasetype == BPCHAROID || targetbasetype == VARCHAROID)
- {
- Oid text_id = TEXTOID;
-
- if (can_coerce_type(1, &exprtype, &text_id, ccontext))
- {
- expr = coerce_type(pstate, expr, exprtype, text_id,
- ccontext, cformat);
- if (targetbasetype != targettype)
- {
- /* need to coerce to domain over char or varchar */
- expr = coerce_to_domain(expr, targetbasetype, targettype,
- cformat);
- }
- else
- {
- /*
- * need a RelabelType if no typmod coercion will be
- * performed
- */
- if (targettypmod < 0)
- expr = (Node *) makeRelabelType((Expr *) expr,
- targettype, -1,
- cformat);
- }
- }
- else
- expr = NULL;
- }
- else
- expr = NULL;
- }
else
expr = NULL;
*
* Copyright (c) 2002-2003, PostgreSQL Global Development Group
*
- * $PostgreSQL: pgsql/src/include/catalog/pg_cast.h,v 1.10 2003/11/29 22:40:58 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/catalog/pg_cast.h,v 1.11 2004/03/15 01:13:41 tgl Exp $
*
* NOTES
* the genbki.sh script reads this file and generates .bki
DATA(insert ( 1043 25 0 i ));
DATA(insert ( 1043 1042 0 i ));
DATA(insert ( 18 25 946 i ));
-DATA(insert ( 18 1042 860 i ));
+DATA(insert ( 18 1042 860 a ));
+DATA(insert ( 18 1043 946 a ));
DATA(insert ( 19 25 406 i ));
-DATA(insert ( 19 1042 408 i ));
-DATA(insert ( 19 1043 1401 i ));
+DATA(insert ( 19 1042 408 a ));
+DATA(insert ( 19 1043 1401 a ));
DATA(insert ( 25 18 944 a ));
+DATA(insert ( 1042 18 944 a ));
+DATA(insert ( 1043 18 944 a ));
DATA(insert ( 25 19 407 i ));
DATA(insert ( 1042 19 409 i ));
DATA(insert ( 1043 19 1400 i ));
DATA(insert ( 1700 25 1688 i ));
DATA(insert ( 25 1700 1686 e ));
+/*
+ * Cross-category casts to and from VARCHAR
+ *
+ * We support all the same casts as for TEXT, but none are implicit.
+ */
+DATA(insert ( 20 1043 1289 a ));
+DATA(insert ( 1043 20 1290 e ));
+DATA(insert ( 21 1043 113 a ));
+DATA(insert ( 1043 21 818 e ));
+DATA(insert ( 23 1043 112 a ));
+DATA(insert ( 1043 23 819 e ));
+DATA(insert ( 26 1043 114 a ));
+DATA(insert ( 1043 26 817 e ));
+DATA(insert ( 1043 650 1714 e ));
+DATA(insert ( 700 1043 841 a ));
+DATA(insert ( 1043 700 839 e ));
+DATA(insert ( 701 1043 840 a ));
+DATA(insert ( 1043 701 838 e ));
+DATA(insert ( 829 1043 752 e ));
+DATA(insert ( 1043 829 767 e ));
+DATA(insert ( 650 1043 730 e ));
+DATA(insert ( 869 1043 730 e ));
+DATA(insert ( 1043 869 1713 e ));
+DATA(insert ( 1082 1043 749 a ));
+DATA(insert ( 1043 1082 748 e ));
+DATA(insert ( 1083 1043 948 a ));
+DATA(insert ( 1043 1083 837 e ));
+DATA(insert ( 1114 1043 2034 a ));
+DATA(insert ( 1043 1114 2022 e ));
+DATA(insert ( 1184 1043 1192 a ));
+DATA(insert ( 1043 1184 1191 e ));
+DATA(insert ( 1186 1043 1193 a ));
+DATA(insert ( 1043 1186 1263 e ));
+DATA(insert ( 1266 1043 939 a ));
+DATA(insert ( 1043 1266 938 e ));
+DATA(insert ( 1700 1043 1688 a ));
+DATA(insert ( 1043 1700 1686 e ));
+
+/*
+ * Cross-category casts to and from BPCHAR
+ *
+ * A function supporting cast to TEXT/VARCHAR can be used for cast to BPCHAR,
+ * but the other direction is okay only if the function treats trailing
+ * blanks as insignificant. So this is a subset of the VARCHAR list.
+ * (Arguably the holdouts should be fixed, but I'm not doing that now...)
+ */
+DATA(insert ( 20 1042 1289 a ));
+DATA(insert ( 1042 20 1290 e ));
+DATA(insert ( 21 1042 113 a ));
+DATA(insert ( 1042 21 818 e ));
+DATA(insert ( 23 1042 112 a ));
+DATA(insert ( 1042 23 819 e ));
+DATA(insert ( 26 1042 114 a ));
+DATA(insert ( 1042 26 817 e ));
+DATA(insert ( 700 1042 841 a ));
+DATA(insert ( 1042 700 839 e ));
+DATA(insert ( 701 1042 840 a ));
+DATA(insert ( 1042 701 838 e ));
+DATA(insert ( 829 1042 752 e ));
+DATA(insert ( 1042 829 767 e ));
+DATA(insert ( 650 1042 730 e ));
+DATA(insert ( 869 1042 730 e ));
+DATA(insert ( 1082 1042 749 a ));
+DATA(insert ( 1042 1082 748 e ));
+DATA(insert ( 1083 1042 948 a ));
+DATA(insert ( 1042 1083 837 e ));
+DATA(insert ( 1114 1042 2034 a ));
+DATA(insert ( 1042 1114 2022 e ));
+DATA(insert ( 1184 1042 1192 a ));
+DATA(insert ( 1042 1184 1191 e ));
+DATA(insert ( 1186 1042 1193 a ));
+DATA(insert ( 1042 1186 1263 e ));
+DATA(insert ( 1266 1042 939 a ));
+DATA(insert ( 1042 1266 938 e ));
+DATA(insert ( 1700 1042 1688 a ));
+DATA(insert ( 1042 1700 1686 e ));
+
#endif /* PG_CAST_H */
-- Look for cast functions that don't have the right signature. The
-- argument and result types in pg_proc must be the same as, or binary
-- compatible with, what it says in pg_cast.
+-- As a special case, we allow casts from CHAR(n) that use functions
+-- declared to take TEXT. This does not pass the binary-coercibility test
+-- because CHAR(n)-to-TEXT normally invokes rtrim(). However, the results
+-- are the same, so long as the function is one that ignores trailing blanks.
SELECT c.*
FROM pg_cast c, pg_proc p
WHERE c.castfunc = p.oid AND
(p.pronargs <> 1
- OR NOT binary_coercible(c.castsource, p.proargtypes[0])
+ OR NOT (binary_coercible(c.castsource, p.proargtypes[0])
+ OR (c.castsource = 'character'::regtype AND
+ p.proargtypes[0] = 'text'::regtype))
OR NOT binary_coercible(p.prorettype, c.casttarget));
castsource | casttarget | castfunc | castcontext
------------+------------+----------+-------------
-- Look for cast functions that don't have the right signature. The
-- argument and result types in pg_proc must be the same as, or binary
-- compatible with, what it says in pg_cast.
+-- As a special case, we allow casts from CHAR(n) that use functions
+-- declared to take TEXT. This does not pass the binary-coercibility test
+-- because CHAR(n)-to-TEXT normally invokes rtrim(). However, the results
+-- are the same, so long as the function is one that ignores trailing blanks.
SELECT c.*
FROM pg_cast c, pg_proc p
WHERE c.castfunc = p.oid AND
(p.pronargs <> 1
- OR NOT binary_coercible(c.castsource, p.proargtypes[0])
+ OR NOT (binary_coercible(c.castsource, p.proargtypes[0])
+ OR (c.castsource = 'character'::regtype AND
+ p.proargtypes[0] = 'text'::regtype))
OR NOT binary_coercible(p.prorettype, c.casttarget));
-- Look for binary compatible casts that do not have the reverse