1 /*-------------------------------------------------------------------------
4 * Asynchronous notification: NOTIFY, LISTEN, UNLISTEN
6 * Portions Copyright (c) 1996-2003, PostgreSQL Global Development Group
7 * Portions Copyright (c) 1994, Regents of the University of California
10 * $PostgreSQL: pgsql/src/backend/commands/async.c,v 1.108 2004/01/27 00:45:26 momjian Exp $
12 *-------------------------------------------------------------------------
15 /*-------------------------------------------------------------------------
16 * New Async Notification Model:
17 * 1. Multiple backends on same machine. Multiple backends listening on
18 * one relation. (Note: "listening on a relation" is not really the
19 * right way to think about it, since the notify names need not have
20 * anything to do with the names of relations actually in the database.
21 * But this terminology is all over the code and docs, and I don't feel
22 * like trying to replace it.)
24 * 2. There is a tuple in relation "pg_listener" for each active LISTEN,
25 * ie, each relname/listenerPID pair. The "notification" field of the
26 * tuple is zero when no NOTIFY is pending for that listener, or the PID
27 * of the originating backend when a cross-backend NOTIFY is pending.
28 * (We skip writing to pg_listener when doing a self-NOTIFY, so the
29 * notification field should never be equal to the listenerPID field.)
31 * 3. The NOTIFY statement itself (routine Async_Notify) just adds the target
32 * relname to a list of outstanding NOTIFY requests. Actual processing
33 * happens if and only if we reach transaction commit. At that time (in
34 * routine AtCommit_Notify) we scan pg_listener for matching relnames.
35 * If the listenerPID in a matching tuple is ours, we just send a notify
36 * message to our own front end. If it is not ours, and "notification"
37 * is not already nonzero, we set notification to our own PID and send a
38 * SIGUSR2 signal to the receiving process (indicated by listenerPID).
39 * BTW: if the signal operation fails, we presume that the listener backend
40 * crashed without removing this tuple, and remove the tuple for it.
42 * 4. Upon receipt of a SIGUSR2 signal, the signal handler can call inbound-
43 * notify processing immediately if this backend is idle (ie, it is
44 * waiting for a frontend command and is not within a transaction block).
45 * Otherwise the handler may only set a flag, which will cause the
46 * processing to occur just before we next go idle.
48 * 5. Inbound-notify processing consists of scanning pg_listener for tuples
49 * matching our own listenerPID and having nonzero notification fields.
50 * For each such tuple, we send a message to our frontend and clear the
51 * notification field. BTW: this routine has to start/commit its own
52 * transaction, since by assumption it is only called from outside any
55 * Although we grab AccessExclusiveLock on pg_listener for any operation,
56 * the lock is never held very long, so it shouldn't cause too much of
57 * a performance problem.
59 * An application that listens on the same relname it notifies will get
60 * NOTIFY messages for its own NOTIFYs. These can be ignored, if not useful,
61 * by comparing be_pid in the NOTIFY message to the application's own backend's
62 * PID. (As of FE/BE protocol 2.0, the backend's PID is provided to the
63 * frontend during startup.) The above design guarantees that notifies from
64 * other backends will never be missed by ignoring self-notifies. Note,
65 * however, that we do *not* guarantee that a separate frontend message will
66 * be sent for every outside NOTIFY. Since there is only room for one
67 * originating PID in pg_listener, outside notifies occurring at about the
68 * same time may be collapsed into a single message bearing the PID of the
69 * first outside backend to perform the NOTIFY.
70 *-------------------------------------------------------------------------
78 #include <netinet/in.h>
80 #include "access/heapam.h"
81 #include "catalog/catname.h"
82 #include "catalog/pg_listener.h"
83 #include "commands/async.h"
84 #include "libpq/libpq.h"
85 #include "libpq/pqformat.h"
86 #include "libpq/pqsignal.h"
87 #include "miscadmin.h"
88 #include "storage/ipc.h"
89 #include "tcop/tcopprot.h"
90 #include "utils/fmgroids.h"
91 #include "utils/ps_status.h"
92 #include "utils/syscache.h"
96 * State for outbound notifies consists of a list of all relnames NOTIFYed
97 * in the current transaction. We do not actually perform a NOTIFY until
98 * and unless the transaction commits. pendingNotifies is NIL if no
99 * NOTIFYs have been done in the current transaction. The List nodes and
100 * referenced strings are all palloc'd in TopTransactionContext.
102 static List *pendingNotifies = NIL;
105 * State for inbound notifies consists of two flags: one saying whether
106 * the signal handler is currently allowed to call ProcessIncomingNotify
107 * directly, and one saying whether the signal has occurred but the handler
108 * was not allowed to call ProcessIncomingNotify at the time.
110 * NB: the "volatile" on these declarations is critical! If your compiler
111 * does not grok "volatile", you'd be best advised to compile this file
112 * with all optimization turned off.
114 static volatile int notifyInterruptEnabled = 0;
115 static volatile int notifyInterruptOccurred = 0;
117 /* True if we've registered an on_shmem_exit cleanup */
118 static bool unlistenExitRegistered = false;
120 bool Trace_notify = false;
123 static void Async_UnlistenAll(void);
124 static void Async_UnlistenOnExit(int code, Datum arg);
125 static void ProcessIncomingNotify(void);
126 static void NotifyMyFrontEnd(char *relname, int32 listenerPID);
127 static bool AsyncExistsPendingNotify(const char *relname);
128 static void ClearPendingNotifies(void);
132 *--------------------------------------------------------------
135 * This is executed by the SQL notify command.
137 * Adds the relation to the list of pending notifies.
138 * Actual notification happens during transaction commit.
139 * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
144 *--------------------------------------------------------------
147 Async_Notify(char *relname)
150 elog(DEBUG1, "Async_Notify(%s)", relname);
152 /* no point in making duplicate entries in the list ... */
153 if (!AsyncExistsPendingNotify(relname))
156 * The name list needs to live until end of transaction, so store
157 * it in the top transaction context.
159 MemoryContext oldcontext;
161 oldcontext = MemoryContextSwitchTo(TopTransactionContext);
163 pendingNotifies = lcons(pstrdup(relname), pendingNotifies);
165 MemoryContextSwitchTo(oldcontext);
170 *--------------------------------------------------------------
173 * This is executed by the SQL listen command.
175 * Register a backend (identified by its Unix PID) as listening
176 * on the specified relation.
182 * pg_listener is updated.
184 *--------------------------------------------------------------
187 Async_Listen(char *relname, int pid)
192 Datum values[Natts_pg_listener];
193 char nulls[Natts_pg_listener];
195 bool alreadyListener = false;
198 elog(DEBUG1, "Async_Listen(%s,%d)", relname, pid);
200 lRel = heap_openr(ListenerRelationName, AccessExclusiveLock);
202 /* Detect whether we are already listening on this relname */
203 scan = heap_beginscan(lRel, SnapshotNow, 0, NULL);
204 while ((tuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
206 Form_pg_listener listener = (Form_pg_listener) GETSTRUCT(tuple);
208 if (listener->listenerpid == pid &&
209 strncmp(NameStr(listener->relname), relname, NAMEDATALEN) == 0)
211 alreadyListener = true;
212 /* No need to scan the rest of the table */
220 heap_close(lRel, AccessExclusiveLock);
225 * OK to insert a new tuple
228 for (i = 0; i < Natts_pg_listener; i++)
231 values[i] = PointerGetDatum(NULL);
235 values[i++] = (Datum) relname;
236 values[i++] = (Datum) pid;
237 values[i++] = (Datum) 0; /* no notifies pending */
239 tuple = heap_formtuple(RelationGetDescr(lRel), values, nulls);
240 simple_heap_insert(lRel, tuple);
242 #ifdef NOT_USED /* currently there are no indexes */
243 CatalogUpdateIndexes(lRel, tuple);
246 heap_freetuple(tuple);
248 heap_close(lRel, AccessExclusiveLock);
251 * now that we are listening, make sure we will unlisten before dying.
253 if (!unlistenExitRegistered)
255 on_shmem_exit(Async_UnlistenOnExit, 0);
256 unlistenExitRegistered = true;
261 *--------------------------------------------------------------
264 * This is executed by the SQL unlisten command.
266 * Remove the backend from the list of listening backends
267 * for the specified relation.
273 * pg_listener is updated.
275 *--------------------------------------------------------------
278 Async_Unlisten(char *relname, int pid)
284 /* Handle specially the `unlisten "*"' command */
285 if ((!relname) || (*relname == '\0') || (strcmp(relname, "*") == 0))
292 elog(DEBUG1, "Async_Unlisten(%s,%d)", relname, pid);
294 lRel = heap_openr(ListenerRelationName, AccessExclusiveLock);
296 scan = heap_beginscan(lRel, SnapshotNow, 0, NULL);
297 while ((tuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
299 Form_pg_listener listener = (Form_pg_listener) GETSTRUCT(tuple);
301 if (listener->listenerpid == pid &&
302 strncmp(NameStr(listener->relname), relname, NAMEDATALEN) == 0)
304 /* Found the matching tuple, delete it */
305 simple_heap_delete(lRel, &tuple->t_self);
308 * We assume there can be only one match, so no need to scan
309 * the rest of the table
316 heap_close(lRel, AccessExclusiveLock);
319 * We do not complain about unlistening something not being listened;
325 *--------------------------------------------------------------
328 * Unlisten all relations for this backend.
330 * This is invoked by UNLISTEN "*" command, and also at backend exit.
336 * pg_listener is updated.
338 *--------------------------------------------------------------
341 Async_UnlistenAll(void)
350 elog(DEBUG1, "Async_UnlistenAll");
352 lRel = heap_openr(ListenerRelationName, AccessExclusiveLock);
353 tdesc = RelationGetDescr(lRel);
355 /* Find and delete all entries with my listenerPID */
357 Anum_pg_listener_pid,
358 BTEqualStrategyNumber, F_INT4EQ,
359 Int32GetDatum(MyProcPid));
360 scan = heap_beginscan(lRel, SnapshotNow, 1, key);
362 while ((lTuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
363 simple_heap_delete(lRel, &lTuple->t_self);
366 heap_close(lRel, AccessExclusiveLock);
370 *--------------------------------------------------------------
371 * Async_UnlistenOnExit
373 * Clean up the pg_listener table at backend exit.
375 * This is executed if we have done any LISTENs in this backend.
376 * It might not be necessary anymore, if the user UNLISTENed everything,
377 * but we don't try to detect that case.
383 * pg_listener is updated if necessary.
385 *--------------------------------------------------------------
388 Async_UnlistenOnExit(int code, Datum arg)
391 * We need to start/commit a transaction for the unlisten, but if
392 * there is already an active transaction we had better abort that one
393 * first. Otherwise we'd end up committing changes that probably
394 * ought to be discarded.
396 AbortOutOfAnyTransaction();
397 /* Now we can do the unlisten */
398 StartTransactionCommand();
400 CommitTransactionCommand();
404 *--------------------------------------------------------------
407 * This is called at transaction commit.
409 * If there are outbound notify requests in the pendingNotifies list,
410 * scan pg_listener for matching tuples, and either signal the other
411 * backend or send a message to our own frontend.
413 * NOTE: we are still inside the current transaction, therefore can
414 * piggyback on its committing of changes.
420 * Tuples in pg_listener that have matching relnames and other peoples'
421 * listenerPIDs are updated with a nonzero notification field.
423 *--------------------------------------------------------------
426 AtCommit_Notify(void)
433 Datum value[Natts_pg_listener];
434 char repl[Natts_pg_listener],
435 nulls[Natts_pg_listener];
437 if (pendingNotifies == NIL)
438 return; /* no NOTIFY statements in this
442 * NOTIFY is disabled if not normal processing mode. This test used to
443 * be in xact.c, but it seems cleaner to do it here.
445 if (!IsNormalProcessingMode())
447 ClearPendingNotifies();
452 elog(DEBUG1, "AtCommit_Notify");
454 /* preset data to update notify column to MyProcPid */
455 nulls[0] = nulls[1] = nulls[2] = ' ';
456 repl[0] = repl[1] = repl[2] = ' ';
457 repl[Anum_pg_listener_notify - 1] = 'r';
458 value[0] = value[1] = value[2] = (Datum) 0;
459 value[Anum_pg_listener_notify - 1] = Int32GetDatum(MyProcPid);
461 lRel = heap_openr(ListenerRelationName, AccessExclusiveLock);
462 tdesc = RelationGetDescr(lRel);
463 scan = heap_beginscan(lRel, SnapshotNow, 0, NULL);
465 while ((lTuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
467 Form_pg_listener listener = (Form_pg_listener) GETSTRUCT(lTuple);
468 char *relname = NameStr(listener->relname);
469 int32 listenerPID = listener->listenerpid;
471 if (!AsyncExistsPendingNotify(relname))
474 if (listenerPID == MyProcPid)
477 * Self-notify: no need to bother with table update. Indeed,
478 * we *must not* clear the notification field in this path, or
479 * we could lose an outside notify, which'd be bad for
480 * applications that ignore self-notify messages.
484 elog(DEBUG1, "AtCommit_Notify: notifying self");
486 NotifyMyFrontEnd(relname, listenerPID);
491 elog(DEBUG1, "AtCommit_Notify: notifying pid %d",
495 * If someone has already notified this listener, we don't
496 * bother modifying the table, but we do still send a SIGUSR2
497 * signal, just in case that backend missed the earlier signal
498 * for some reason. It's OK to send the signal first, because
499 * the other guy can't read pg_listener until we unlock it.
501 if (pqkill(listenerPID, SIGUSR2) < 0)
504 * Get rid of pg_listener entry if it refers to a PID that
505 * no longer exists. Presumably, that backend crashed
506 * without deleting its pg_listener entries. This code
507 * used to only delete the entry if errno==ESRCH, but as
508 * far as I can see we should just do it for any failure
509 * (certainly at least for EPERM too...)
511 simple_heap_delete(lRel, &lTuple->t_self);
513 else if (listener->notification == 0)
515 ItemPointerData ctid;
518 rTuple = heap_modifytuple(lTuple, lRel,
521 * We cannot use simple_heap_update here because the tuple
522 * could have been modified by an uncommitted transaction;
523 * specifically, since UNLISTEN releases exclusive lock on
524 * the table before commit, the other guy could already have
525 * tried to unlisten. There are no other cases where we
526 * should be able to see an uncommitted update or delete.
527 * Therefore, our response to a HeapTupleBeingUpdated result
528 * is just to ignore it. We do *not* wait for the other
529 * guy to commit --- that would risk deadlock, and we don't
530 * want to block while holding the table lock anyway for
531 * performance reasons. We also ignore HeapTupleUpdated,
532 * which could occur if the other guy commits between our
533 * heap_getnext and heap_update calls.
535 result = heap_update(lRel, &lTuple->t_self, rTuple,
537 GetCurrentCommandId(), SnapshotAny,
538 false /* no wait for commit */);
541 case HeapTupleSelfUpdated:
542 /* Tuple was already updated in current command? */
543 elog(ERROR, "tuple already updated by self");
546 case HeapTupleMayBeUpdated:
547 /* done successfully */
549 #ifdef NOT_USED /* currently there are no indexes */
550 CatalogUpdateIndexes(lRel, rTuple);
554 case HeapTupleBeingUpdated:
555 /* ignore uncommitted tuples */
558 case HeapTupleUpdated:
559 /* ignore just-committed tuples */
563 elog(ERROR, "unrecognized heap_update status: %u",
574 * We do NOT release the lock on pg_listener here; we need to hold it
575 * until end of transaction (which is about to happen, anyway) to
576 * ensure that notified backends see our tuple updates when they look.
577 * Else they might disregard the signal, which would make the
578 * application programmer very unhappy.
580 heap_close(lRel, NoLock);
582 ClearPendingNotifies();
585 elog(DEBUG1, "AtCommit_Notify: done");
589 *--------------------------------------------------------------
592 * This is called at transaction abort.
594 * Gets rid of pending outbound notifies that we would have executed
595 * if the transaction got committed.
600 *--------------------------------------------------------------
605 ClearPendingNotifies();
609 *--------------------------------------------------------------
610 * Async_NotifyHandler
612 * This is the signal handler for SIGUSR2.
614 * If we are idle (notifyInterruptEnabled is set), we can safely invoke
615 * ProcessIncomingNotify directly. Otherwise, just set a flag
623 *--------------------------------------------------------------
626 Async_NotifyHandler(SIGNAL_ARGS)
628 int save_errno = errno;
631 * Note: this is a SIGNAL HANDLER. You must be very wary what you do
632 * here. Some helpful soul had this routine sprinkled with TPRINTFs,
633 * which would likely lead to corruption of stdio buffers if they were
637 /* Don't joggle the elbow of proc_exit */
638 if (proc_exit_inprogress)
641 if (notifyInterruptEnabled)
643 bool save_ImmediateInterruptOK = ImmediateInterruptOK;
646 * We may be called while ImmediateInterruptOK is true; turn it
647 * off while messing with the NOTIFY state. (We would have to
648 * save and restore it anyway, because PGSemaphore operations
649 * inside ProcessIncomingNotify() might reset it.)
651 ImmediateInterruptOK = false;
654 * I'm not sure whether some flavors of Unix might allow another
655 * SIGUSR2 occurrence to recursively interrupt this routine. To
656 * cope with the possibility, we do the same sort of dance that
657 * EnableNotifyInterrupt must do --- see that routine for
660 notifyInterruptEnabled = 0; /* disable any recursive signal */
661 notifyInterruptOccurred = 1; /* do at least one iteration */
664 notifyInterruptEnabled = 1;
665 if (!notifyInterruptOccurred)
667 notifyInterruptEnabled = 0;
668 if (notifyInterruptOccurred)
670 /* Here, it is finally safe to do stuff. */
672 elog(DEBUG1, "Async_NotifyHandler: perform async notify");
674 ProcessIncomingNotify();
677 elog(DEBUG1, "Async_NotifyHandler: done");
682 * Restore ImmediateInterruptOK, and check for interrupts if
685 ImmediateInterruptOK = save_ImmediateInterruptOK;
686 if (save_ImmediateInterruptOK)
687 CHECK_FOR_INTERRUPTS();
692 * In this path it is NOT SAFE to do much of anything, except
695 notifyInterruptOccurred = 1;
702 * --------------------------------------------------------------
703 * EnableNotifyInterrupt
705 * This is called by the PostgresMain main loop just before waiting
706 * for a frontend command. If we are truly idle (ie, *not* inside
707 * a transaction block), then process any pending inbound notifies,
708 * and enable the signal handler to process future notifies directly.
710 * NOTE: the signal handler starts out disabled, and stays so until
711 * PostgresMain calls this the first time.
712 * --------------------------------------------------------------
715 EnableNotifyInterrupt(void)
717 if (IsTransactionOrTransactionBlock())
718 return; /* not really idle */
721 * This code is tricky because we are communicating with a signal
722 * handler that could interrupt us at any point. If we just checked
723 * notifyInterruptOccurred and then set notifyInterruptEnabled, we
724 * could fail to respond promptly to a signal that happens in between
725 * those two steps. (A very small time window, perhaps, but Murphy's
726 * Law says you can hit it...) Instead, we first set the enable flag,
727 * then test the occurred flag. If we see an unserviced interrupt has
728 * occurred, we re-clear the enable flag before going off to do the
729 * service work. (That prevents re-entrant invocation of
730 * ProcessIncomingNotify() if another interrupt occurs.) If an
731 * interrupt comes in between the setting and clearing of
732 * notifyInterruptEnabled, then it will have done the service work and
733 * left notifyInterruptOccurred zero, so we have to check again after
734 * clearing enable. The whole thing has to be in a loop in case
735 * another interrupt occurs while we're servicing the first. Once we
736 * get out of the loop, enable is set and we know there is no
737 * unserviced interrupt.
739 * NB: an overenthusiastic optimizing compiler could easily break this
740 * code. Hopefully, they all understand what "volatile" means these
745 notifyInterruptEnabled = 1;
746 if (!notifyInterruptOccurred)
748 notifyInterruptEnabled = 0;
749 if (notifyInterruptOccurred)
752 elog(DEBUG1, "EnableNotifyInterrupt: perform async notify");
754 ProcessIncomingNotify();
757 elog(DEBUG1, "EnableNotifyInterrupt: done");
763 * --------------------------------------------------------------
764 * DisableNotifyInterrupt
766 * This is called by the PostgresMain main loop just after receiving
767 * a frontend command. Signal handler execution of inbound notifies
768 * is disabled until the next EnableNotifyInterrupt call.
769 * --------------------------------------------------------------
772 DisableNotifyInterrupt(void)
774 notifyInterruptEnabled = 0;
778 * --------------------------------------------------------------
779 * ProcessIncomingNotify
781 * Deal with arriving NOTIFYs from other backends.
782 * This is called either directly from the SIGUSR2 signal handler,
783 * or the next time control reaches the outer idle loop.
784 * Scan pg_listener for arriving notifies, report them to my front end,
785 * and clear the notification field in pg_listener until next time.
787 * NOTE: since we are outside any transaction, we must create our own.
792 * --------------------------------------------------------------
795 ProcessIncomingNotify(void)
803 Datum value[Natts_pg_listener];
804 char repl[Natts_pg_listener],
805 nulls[Natts_pg_listener];
808 elog(DEBUG1, "ProcessIncomingNotify");
810 set_ps_display("async_notify");
812 notifyInterruptOccurred = 0;
814 StartTransactionCommand();
816 lRel = heap_openr(ListenerRelationName, AccessExclusiveLock);
817 tdesc = RelationGetDescr(lRel);
819 /* Scan only entries with my listenerPID */
821 Anum_pg_listener_pid,
822 BTEqualStrategyNumber, F_INT4EQ,
823 Int32GetDatum(MyProcPid));
824 scan = heap_beginscan(lRel, SnapshotNow, 1, key);
826 /* Prepare data for rewriting 0 into notification field */
827 nulls[0] = nulls[1] = nulls[2] = ' ';
828 repl[0] = repl[1] = repl[2] = ' ';
829 repl[Anum_pg_listener_notify - 1] = 'r';
830 value[0] = value[1] = value[2] = (Datum) 0;
831 value[Anum_pg_listener_notify - 1] = Int32GetDatum(0);
833 while ((lTuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
835 Form_pg_listener listener = (Form_pg_listener) GETSTRUCT(lTuple);
836 char *relname = NameStr(listener->relname);
837 int32 sourcePID = listener->notification;
841 /* Notify the frontend */
844 elog(DEBUG1, "ProcessIncomingNotify: received %s from %d",
845 relname, (int) sourcePID);
847 NotifyMyFrontEnd(relname, sourcePID);
849 * Rewrite the tuple with 0 in notification column.
851 * simple_heap_update is safe here because no one else would
852 * have tried to UNLISTEN us, so there can be no uncommitted
855 rTuple = heap_modifytuple(lTuple, lRel, value, nulls, repl);
856 simple_heap_update(lRel, &lTuple->t_self, rTuple);
858 #ifdef NOT_USED /* currently there are no indexes */
859 CatalogUpdateIndexes(lRel, rTuple);
866 * We do NOT release the lock on pg_listener here; we need to hold it
867 * until end of transaction (which is about to happen, anyway) to
868 * ensure that other backends see our tuple updates when they look.
869 * Otherwise, a transaction started after this one might mistakenly
870 * think it doesn't need to send this backend a new NOTIFY.
872 heap_close(lRel, NoLock);
874 CommitTransactionCommand();
877 * Must flush the notify messages to ensure frontend gets them
882 set_ps_display("idle");
885 elog(DEBUG1, "ProcessIncomingNotify: done");
889 * Send NOTIFY message to my front end.
892 NotifyMyFrontEnd(char *relname, int32 listenerPID)
894 if (whereToSendOutput == Remote)
898 pq_beginmessage(&buf, 'A');
899 pq_sendint(&buf, listenerPID, sizeof(int32));
900 pq_sendstring(&buf, relname);
901 if (PG_PROTOCOL_MAJOR(FrontendProtocol) >= 3)
903 /* XXX Add parameter string here later */
904 pq_sendstring(&buf, "");
909 * NOTE: we do not do pq_flush() here. For a self-notify, it will
910 * happen at the end of the transaction, and for incoming notifies
911 * ProcessIncomingNotify will do it after finding all the
916 elog(INFO, "NOTIFY for %s", relname);
919 /* Does pendingNotifies include the given relname? */
921 AsyncExistsPendingNotify(const char *relname)
925 foreach(p, pendingNotifies)
927 /* Use NAMEDATALEN for relname comparison. DZ - 26-08-1996 */
928 if (strncmp((const char *) lfirst(p), relname, NAMEDATALEN) == 0)
935 /* Clear the pendingNotifies list. */
937 ClearPendingNotifies(void)
940 * We used to have to explicitly deallocate the list members and
941 * nodes, because they were malloc'd. Now, since we know they are
942 * palloc'd in TopTransactionContext, we need not do that --- they'll
943 * go away automatically at transaction exit. We need only reset the
946 pendingNotifies = NIL;