(2014/09/01 20:15), Etsuro Fujita wrote:
> While working on [1], I've found that postgres_fdw behaves oddly:
> 
> postgres=# create foreign table ft (a int) server loopback options
> (table_name 't');
> CREATE FOREIGN TABLE
> postgres=# select tableoid, * from ft;
>   tableoid | a
> ----------+---
>      16400 | 1
> (1 row)
> 
> postgres=# select tableoid, * from ft where tableoid = 16400;
>   tableoid | a
> ----------+---
> (0 rows)

> I think that one simple way of fixing such issues would be
> to consider unsafe to send to the remote a qual that contains any system
> columns.

I noticed the previous patch has overdone it.  Attached is an updated
version of the patch.

Thanks,

PS:
> [1] https://commitfest.postgresql.org/action/patch_view?id=1386

I'll update the patch in [1] on top of this version.

Best regards,
Etsuro Fujita
*** a/contrib/postgres_fdw/deparse.c
--- b/contrib/postgres_fdw/deparse.c
***************
*** 252,257 **** foreign_expr_walker(Node *node,
--- 252,263 ----
  				if (var->varno == glob_cxt->foreignrel->relid &&
  					var->varlevelsup == 0)
  				{
+ 					/*
+ 					 * System columns can't be sent to remote.
+ 					 */
+ 					if (var->varattno < 0)
+ 						return false;
+ 
  					/* Var belongs to foreign table */
  					collation = var->varcollid;
  					state = OidIsValid(collation) ? FDW_COLLATE_SAFE : FDW_COLLATE_NONE;
*** a/src/backend/optimizer/plan/createplan.c
--- b/src/backend/optimizer/plan/createplan.c
***************
*** 20,25 ****
--- 20,26 ----
  #include <math.h>
  
  #include "access/skey.h"
+ #include "access/sysattr.h"
  #include "catalog/pg_class.h"
  #include "foreign/fdwapi.h"
  #include "miscadmin.h"
***************
*** 1945,1950 **** create_foreignscan_plan(PlannerInfo *root, ForeignPath *best_path,
--- 1946,1953 ----
  	RelOptInfo *rel = best_path->path.parent;
  	Index		scan_relid = rel->relid;
  	RangeTblEntry *rte;
+ 	Bitmapset  *attrs_used = NULL;
+ 	ListCell   *lc;
  	int			i;
  
  	/* it should be a base rel... */
***************
*** 1993,2008 **** create_foreignscan_plan(PlannerInfo *root, ForeignPath *best_path,
  	 * bit of a kluge and might go away someday, so we intentionally leave it
  	 * out of the API presented to FDWs.
  	 */
  	scan_plan->fsSystemCol = false;
  	for (i = rel->min_attr; i < 0; i++)
  	{
! 		if (!bms_is_empty(rel->attr_needed[i - rel->min_attr]))
  		{
  			scan_plan->fsSystemCol = true;
  			break;
  		}
  	}
  
  	return scan_plan;
  }
  
--- 1996,2030 ----
  	 * bit of a kluge and might go away someday, so we intentionally leave it
  	 * out of the API presented to FDWs.
  	 */
+ 
+ 	/*
+ 	 * Add all the attributes needed for joins or final output.  Note: we must
+ 	 * look at reltargetlist, not the attr_needed data, because attr_needed
+ 	 * isn't computed for inheritance child rels.
+ 	 */
+ 	pull_varattnos((Node *) rel->reltargetlist, rel->relid, &attrs_used);
+ 
+ 	/* Add all the attributes used by restriction clauses. */
+ 	foreach(lc, rel->baserestrictinfo)
+ 	{
+ 		RestrictInfo *rinfo = (RestrictInfo *) lfirst(lc);
+ 
+ 		pull_varattnos((Node *) rinfo->clause, rel->relid, &attrs_used);
+ 	}
+ 
+ 	/* Are any system columns requested from rel? */
  	scan_plan->fsSystemCol = false;
  	for (i = rel->min_attr; i < 0; i++)
  	{
! 		if (bms_is_member(i - FirstLowInvalidHeapAttributeNumber, attrs_used))
  		{
  			scan_plan->fsSystemCol = true;
  			break;
  		}
  	}
  
+ 	bms_free(attrs_used);
+ 
  	return scan_plan;
  }
  
-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Reply via email to