database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-11-29 23:43:17 +03:00
Code Activity

Be more tense about not creating tuplestores with randomAccess = true unless

Browse Source 
backwards scan could actually happen.  In particular, pass a flag to
materialize-mode SRFs that tells them whether they need to require random
access.  In passing, also suppress unneeded backward-scan overhead for a
Portal's holdStore tuplestore.  Per my proposal about reducing I/O costs for
tuplestores.
This commit is contained in:
Tom Lane
2008-10-29 00:00:39 +00:00
parent e3e3d2a789
commit 05bba3d176
11 changed files with 67 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/backend/utils/fmgr/README
View File

@@ -1,4 +1,4 @@
$PostgreSQL: pgsql/src/backend/utils/fmgr/README,v 1.14 2008/10/28 22:02:05 tgl Exp $
$PostgreSQL: pgsql/src/backend/utils/fmgr/README,v 1.15 2008/10/29 00:00:38 tgl Exp $
Function Manager
================
@@ -432,6 +432,10 @@ function is called in). The function stores pointers to the Tuplestore and
TupleDesc into ReturnSetInfo, sets returnMode to indicate materialize mode,
and returns null. isDone is not used and should be left at ExprSingleResult.
The Tuplestore must be created with randomAccess = true if
SFRM_Materialize_Random is set in allowedModes, but it can (and preferably
should) be created with randomAccess = false if not.
If available, the expected tuple descriptor is passed in ReturnSetInfo;
in other contexts the expectedDesc field will be NULL. The function need
not pay attention to expectedDesc, but it may be useful in special cases.

18
src/backend/utils/mmgr/portalmem.c
View File

@@ -12,7 +12,7 @@
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/utils/mmgr/portalmem.c,v 1.111 2008/07/18 20:26:06 tgl Exp $
* $PostgreSQL: pgsql/src/backend/utils/mmgr/portalmem.c,v 1.112 2008/10/29 00:00:38 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@@ -354,11 +354,17 @@ PortalCreateHoldStore(Portal portal)
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
/* Create the tuple store, selecting cross-transaction temp files. */
/*
* Create the tuple store, selecting cross-transaction temp files, and
* enabling random access only if cursor requires scrolling.
*
* XXX: Should maintenance_work_mem be used for the portal size?
*/
oldcxt = MemoryContextSwitchTo(portal->holdContext);
/* XXX: Should maintenance_work_mem be used for the portal size? */
portal->holdStore = tuplestore_begin_heap(true, true, work_mem);
portal->holdStore =
tuplestore_begin_heap(portal->cursorOptions & CURSOR_OPT_SCROLL,
true, work_mem);
MemoryContextSwitchTo(oldcxt);
}
@@ -913,7 +919,9 @@ pg_cursor(PG_FUNCTION_ARGS)
* We put all the tuples into a tuplestore in one scan of the hashtable.
* This avoids any issue of the hashtable possibly changing between calls.
*/
tupstore = tuplestore_begin_heap(true, false, work_mem);
tupstore =
tuplestore_begin_heap(rsinfo->allowedModes & SFRM_Materialize_Random,
false, work_mem);
hash_seq_init(&hash_seq, PortalHashTable);
while ((hentry = hash_seq_search(&hash_seq)) != NULL)