Re: bgwriter, inherited temp tables TODO items?

"Thomas F. O'Connell" <tfo(at)sitening(dot)com> writes:
> Tom seems to come pretty close to a TODO item in his analysis in my
> opinion. Something like:

> "Make the planner ignore temp tables of other backends when expanding
> an inheritance list."

I've done this in CVS tip. I'm not sure whether it should be considered
a backpatchable bug fix, though.

If you want to apply the patch locally, it's attached --- should work
fine in 8.0, but I'm not sure about 7.4 or earlier, which have slightly
different logic here.

regards, tom lane

*** src/backend/optimizer/prep/prepunion.c.orig Thu Jul 28 18:27:00 2005
--- src/backend/optimizer/prep/prepunion.c Tue Aug 2 16:21:41 2005
***************
*** 22,27 ****
--- 22,28 ----


#include "access/heapam.h"
+ #include "catalog/namespace.h"
#include "catalog/pg_type.h"
#include "nodes/makefuncs.h"
#include "optimizer/clauses.h"
***************
*** 808,813 ****
--- 809,824 ----
Index childRTindex;

/*
+ * It is possible that the parent table has children that are
+ * temp tables of other backends. We cannot safely access such
+ * tables (because of buffering issues), and the best thing to do
+ * seems to be to silently ignore them.
+ */
+ if (childOID != parentOID &&
+ isOtherTempNamespace(get_rel_namespace(childOID)))
+ continue;
+
+ /*
* Build an RTE for the child, and attach to query's rangetable
* list. We copy most fields of the parent's RTE, but replace
* relation OID, and set inh = false.
***************
*** 818,823 ****
--- 829,845 ----
parse->rtable = lappend(parse->rtable, childrte);
childRTindex = list_length(parse->rtable);
inhRTIs = lappend_int(inhRTIs, childRTindex);
+ }
+
+ /*
+ * If all the children were temp tables, pretend it's a non-inheritance
+ * situation. The duplicate RTE we added for the parent table is harmless.
+ */
+ if (list_length(inhRTIs) < 2)
+ {
+ /* Clear flag to save repeated tests if called again */
+ rte->inh = false;
+ return NIL;
}

/*