Re: 8.4beta[12] set returning function fails -- was O.K. with 8.3

2009/6/11 J. Greg Davidson <jgd(at)well(dot)com>:
> Dear PostgreSQL Hackers,
>
> Through PostgreSQL 8.3, both of the following functions worked, using
>  generate_series(array_lower($1, 1), array_upper($1, 1)) i
> instead of generate_subscripts($1, 1).
>
> With PostgreSQL 8.4, both are accepted, but only the second one works,
> regardless of whether I use generate_subscripts or the old way.  The error
> is shown.  What's going on?
>
> Thanks,
>
> _Greg
>
> CREATE OR REPLACE
> FUNCTION array_to_set(ANYARRAY) RETURNS SETOF RECORD AS $$
>  SELECT i AS "index", $1[i] AS "value" FROM generate_subscripts($1, 1) i
> $$ LANGUAGE SQL STRICT IMMUTABLE;
> COMMENT ON FUNCTION array_to_set(ANYARRAY) IS
> 'returns the array as a set of RECORD(index, value) pairs';
>
> SELECT array_to_set(ARRAY['one', 'two']);
>
> -- BREAKS IN PG 8.4 beta1 & beta2, vis:
> --
> --      ERROR:  0A000: set-valued function called in context that cannot accept
> a set
> --      CONTEXT:  SQL function "array_to_set" during startup
> --      LOCATION:  fmgr_sql, functions.c:644

I grep'ed HEAD and found the following change helps this:

diff --git a/src/backend/executor/functions.c b/src/backend/executor/functions.c
index c0261fe..0882671 100644
--- a/src/backend/executor/functions.c
+++ b/src/backend/executor/functions.c
@@ -637,8 +637,7 @@ fmgr_sql(PG_FUNCTION_ARGS)
*/
if (!rsi || !IsA(rsi, ReturnSetInfo) ||
(rsi->allowedModes & SFRM_ValuePerCall) == 0 ||
- (rsi->allowedModes & SFRM_Materialize) == 0 ||
- rsi->expectedDesc == NULL)
+ (rsi->allowedModes & SFRM_Materialize) == 0)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("set-valued function
called in context that cannot accept a set")));

I am not completely sure but rsi->expectedDesc check seems not needed
as before. All regression tests passed.

Regards,

--
Hitoshi Harada