Get rid of postgres_fdw's assumption that remote type OIDs match ours.

The only place we depended on that was in sending numeric type OIDs in
PQexecParams; but we can replace that usage with explicitly casting
each Param symbol in the query string, so that the types are specified
to the remote by name not OID.  This makes no immediate difference but
will be essential if we ever hope to support use of non-builtin types.
This commit is contained in:
Tom Lane 2013-02-22 06:36:09 -05:00
parent 6c4f6664b2
commit 5fd386bb31
3 changed files with 32 additions and 13 deletions

View File

@ -331,12 +331,10 @@ foreign_expr_walker(Node *node, foreign_expr_cxt *context)
* function or type defined in the information_schema.
*
* Our constraints for dealing with types are tighter than they are for
* functions or operators: we want to accept only types that are in pg_catalog
* (else format_type might incorrectly fail to schema-qualify their names),
* and we want to be sure that the remote server will use the same OID as
* we do (since we must transmit a numeric type OID when passing a value of
* the type as a query parameter). Both of these are reasons to reject
* objects created post-bootstrap.
* functions or operators: we want to accept only types that are in pg_catalog,
* else format_type might incorrectly fail to schema-qualify their names.
* (This could be fixed with some changes to format_type, but for now there's
* no need.) Thus we must exclude information_schema types.
*
* XXX there is a problem with this, which is that the set of built-in
* objects expands over time. Something that is built-in to us might not
@ -794,12 +792,20 @@ deparseConst(StringInfo buf, Const *node, PlannerInfo *root)
* We don't need to renumber the parameter ID, because the executor functions
* in postgres_fdw.c preserve the numbering of PARAM_EXTERN Params.
* (This might change soon.)
*
* Note: we label the Param's type explicitly rather than relying on
* transmitting a numeric type OID in PQexecParams(). This allows us to
* avoid assuming that types have the same OIDs on the remote side as they
* do locally --- they need only have the same names.
*/
static void
deparseParam(StringInfo buf, Param *node, PlannerInfo *root)
{
Assert(node->paramkind == PARAM_EXTERN);
appendStringInfo(buf, "$%d", node->paramid);
appendStringInfo(buf, "::%s",
format_type_with_typemod(node->paramtype,
node->paramtypmod));
}
/*

View File

@ -558,11 +558,11 @@ EXPLAIN (VERBOSE, COSTS false) EXECUTE st4(1);
-- once we try it enough times, should switch to generic plan
EXPLAIN (VERBOSE, COSTS false) EXECUTE st4(1);
QUERY PLAN
----------------------------------------------------------------------------------------------
QUERY PLAN
-------------------------------------------------------------------------------------------------------
Foreign Scan on public.ft1 t1
Output: c1, c2, c3, c4, c5, c6, c7, c8
Remote SQL: SELECT "C 1", c2, c3, c4, c5, c6, c7, c8 FROM "S 1"."T 1" WHERE (("C 1" = $1))
Remote SQL: SELECT "C 1", c2, c3, c4, c5, c6, c7, c8 FROM "S 1"."T 1" WHERE (("C 1" = $1::integer))
(3 rows)
-- value of $1 should not be sent to remote
@ -613,12 +613,12 @@ EXPLAIN (VERBOSE, COSTS false) EXECUTE st5('foo', 1);
(4 rows)
EXPLAIN (VERBOSE, COSTS false) EXECUTE st5('foo', 1);
QUERY PLAN
----------------------------------------------------------------------------------------------
QUERY PLAN
-------------------------------------------------------------------------------------------------------
Foreign Scan on public.ft1 t1
Output: c1, c2, c3, c4, c5, c6, c7, c8
Filter: (t1.c8 = $1)
Remote SQL: SELECT "C 1", c2, c3, c4, c5, c6, c7, c8 FROM "S 1"."T 1" WHERE (("C 1" = $2))
Remote SQL: SELECT "C 1", c2, c3, c4, c5, c6, c7, c8 FROM "S 1"."T 1" WHERE (("C 1" = $2::integer))
(4 rows)
EXECUTE st5('foo', 1);

View File

@ -901,11 +901,24 @@ create_cursor(ForeignScanState *node)
if (!OidIsValid(prm->ptype) && params->paramFetch != NULL)
params->paramFetch(params, paramno);
/*
* Force the remote server to infer a type for this parameter.
* Since we explicitly cast every parameter (see deparse.c), the
* "inference" is trivial and will produce the desired result.
* This allows us to avoid assuming that the remote server has the
* same OIDs we do for the parameters' types.
*
* We'd not need to pass a type array to PQexecParams at all,
* except that there may be unused holes in the array, which
* will have to be filled with something or the remote server will
* complain. We arbitrarily set them to INT4OID earlier.
*/
types[paramno - 1] = InvalidOid;
/*
* Get string representation of each parameter value by invoking
* type-specific output function, unless the value is null.
*/
types[paramno - 1] = prm->ptype;
if (prm->isnull)
values[paramno - 1] = NULL;
else