On Tue, 2007-10-16 at 00:34 -0400, Tom Lane wrote: > Seems like this needs more comments about what's happening, rather > than less ...
Fair point. > Also, it looks to me like the plan node's own resultslot might never be > assigned to at all, when the subplan returns zero rows. Does this > corner case still work correctly? ISTM the node's own result slot wouldn't be assigned to in this case regardless: (nodeMaterial.c, circa 116) outerslot = ExecProcNode(outerNode); if (TupIsNull(outerslot)) { node->eof_underlying = true; return NULL; } There's no requirement that we must assign to the result slot, AFAICS. -Neil ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match