1
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
8
CONTENT="Modular DocBook HTML Stylesheet Version 1.79"><LINK
10
HREF="mailto:pgsql-docs@postgresql.org"><LINK
12
TITLE="PostgreSQL 9.3.5 Documentation"
13
HREF="index.html"><LINK
16
HREF="release.html"><LINK
19
HREF="release-9-3.html"><LINK
22
HREF="release-9-2-8.html"><LINK
25
HREF="stylesheet.css"><META
26
HTTP-EQUIV="Content-Type"
27
CONTENT="text/html; charset=ISO-8859-1"><META
29
CONTENT="2014-07-21T19:26:29"></HEAD
35
SUMMARY="Header navigation table"
47
>PostgreSQL 9.3.5 Documentation</A
57
HREF="release-9-3.html"
74
>Appendix E. Release Notes</TD
81
HREF="release-9-2-8.html"
96
>E.7. Release 9.2.9</A
109
> This release contains a variety of fixes from 9.2.8.
110
For information about new features in the 9.2 major release, see
112
HREF="release-9-2.html"
122
>E.7.1. Migration to Version 9.2.9</A
125
> A dump/restore is not required for those running 9.2.X.
128
> However, this release corrects an index corruption problem in some GiST
129
indexes. See the first changelog entry below to find out whether your
130
installation has been affected and what steps you should take if so.
133
> Also, if you are upgrading from a version earlier than 9.2.6,
135
HREF="release-9-2-6.html"
153
> Correctly initialize padding bytes in <TT
155
>contrib/btree_gist</TT
160
> columns (Heikki Linnakangas)
163
> This error could result in incorrect query results due to values that
164
should compare equal not being seen as equal.
165
Users with GiST indexes on <TT
175
> those indexes after installing this
181
> Protect against torn pages when deleting GIN list pages (Heikki
185
> This fix prevents possible index corruption if a system crash occurs
186
while the page update is being written to disk.
191
> Don't clear the right-link of a GiST index page while replaying
192
updates from WAL (Heikki Linnakangas)
195
> This error could lead to transiently wrong answers from GiST index
196
scans performed in Hot Standby.
201
> Fix corner-case infinite loop during insertion into an SP-GiST text
207
> Fix feedback status when <A
208
HREF="runtime-config-replication.html#GUC-HOT-STANDBY-FEEDBACK"
209
>hot_standby_feedback</A
211
turned off on-the-fly (Simon Riggs)
216
> Fix possibly-incorrect cache invalidation during nested calls
219
>ReceiveSharedInvalidMessages</CODE
225
> Fix planner's mishandling of nested PlaceHolderVars generated in
226
nested-nestloop plans (Tom Lane)
229
> This oversight could result in <SPAN
231
>"variable not found in subplan
233
> errors, or in silently wrong query results.
240
>"could not find pathkey item to sort"</SPAN
245
> over subqueries reading from tables with
246
inheritance children (Tom Lane)
251
> Don't assume a subquery's output is unique if there's a set-returning
252
function in its targetlist (David Rowley)
255
> This oversight could lead to misoptimization of constructs
258
>WHERE x IN (SELECT y, generate_series(1,10) FROM t GROUP
265
> Improve planner to drop constant-NULL inputs
272
> when possible (Tom Lane)
275
> This change fixes some cases where the more aggressive parameter
276
substitution done by 9.2 and later can lead to a worse plan than
277
older versions produced.
282
> Fix identification of input type category in <CODE
286
and friends (Tom Lane)
289
> This is known to have led to inadequate quoting of <TT
296
> result, and there may have been wrong
297
results for other data types as well.
302
> Fix failure to detoast fields in composite elements of structured
306
> This corrects cases where TOAST pointers could be copied into other
307
tables without being dereferenced. If the original data is later
308
deleted, it would lead to errors like <SPAN
310
>"missing chunk number 0
311
for toast value ..."</SPAN
312
> when the now-dangling pointer is used.
319
>"record type has not been registered"</SPAN
321
whole-row references to the output of Append plan nodes (Tom Lane)
326
> Fix possible crash when invoking a user-defined function while
327
rewinding a cursor (Tom Lane)
332
> Fix query-lifespan memory leak while evaluating the arguments for a
341
> Fix session-lifespan memory leaks in regular-expression processing
342
(Tom Lane, Arthur O'Dwyer, Greg Stark)
347
> Fix data encoding error in <TT
355
> Prevent foreign tables from being created with OIDS
357
HREF="runtime-config-compatible.html#GUC-DEFAULT-WITH-OIDS"
358
>default_with_oids</A
365
> Fix liveness checks for rows that were inserted in the current
366
transaction and then deleted by a now-rolled-back subtransaction
370
> This could cause problems (at least spurious warnings, and at worst an
371
infinite loop) if <TT
378
done later in the same transaction.
385
>pg_stat_activity</TT
392
>PREPARE TRANSACTION</TT
399
>, the originating session is no longer in
400
a transaction, so it should not continue to display a transaction
409
> to not fail for text search objects
410
(Álvaro Herrera)
415
> Block signals during postmaster startup (Tom Lane)
418
> This ensures that the postmaster will properly clean up after itself
419
if, for example, it receives <SPAN
428
> Fix client host name lookup when processing <TT
432
entries that specify host names instead of IP addresses (Tom Lane)
435
> Ensure that reverse-DNS lookup failures are reported, instead of just
436
silently not matching such entries. Also ensure that we make only
437
one reverse-DNS lookup attempt per connection, not one per host name
438
entry, which is what previously happened if the lookup attempts failed.
443
> Allow the root user to use <TT
445
>postgres -C variable</TT
449
>postgres --describe-config</TT
453
> The prohibition on starting the server as root does not need to extend
454
to these operations, and relaxing it prevents failure
463
> Secure Unix-domain sockets of temporary postmasters started during
470
> Any local user able to access the socket file could connect as the
471
server's bootstrap superuser, then proceed to execute arbitrary code as
472
the operating-system user running the test, as we previously noted in
473
CVE-2014-0067. This change defends against that risk by placing the
474
server's socket in a temporary, mode 0700 subdirectory
478
>. The hazard remains however on platforms where
479
Unix sockets are not supported, notably Windows, because then the
480
temporary postmaster must accept local TCP connections.
483
> A useful side effect of this change is to simplify
487
> testing in builds that
490
>DEFAULT_PGSOCKET_DIR</TT
491
>. Popular non-default values
494
>/var/run/postgresql</TT
495
> are often not writable by the
496
build user, requiring workarounds that will no longer be necessary.
501
> Fix tablespace creation WAL replay to work on Windows (MauMau)
506
> Fix detection of socket creation failures on Windows (Bruce Momjian)
511
> On Windows, allow new sessions to absorb values of PGC_BACKEND
512
parameters (such as <A
513
HREF="runtime-config-logging.html#GUC-LOG-CONNECTIONS"
516
configuration file (Amit Kapila)
519
> Previously, if such a parameter were changed in the file post-startup,
520
the change would have no effect.
525
> Properly quote executable path names on Windows (Nikhil Deshpande)
528
> This oversight could cause <SPAN
535
> to fail on Windows, if the installation
536
path contained both spaces and <TT
544
> Fix linking of <SPAN
550
> The method we previously used can fail with the Python library
551
supplied by Xcode 5.0 and later.
556
> Avoid buffer bloat in <SPAN
560
consistently sends data faster than the client can absorb it
561
(Shin-ichi Morita, Tom Lane)
567
> could be coerced into enlarging its input buffer
568
until it runs out of memory (which would be reported misleadingly
571
>"lost synchronization with server"</SPAN
573
circumstances it's quite far-fetched that data could be continuously
574
transmitted more quickly than the <CODE
578
absorb it, but this has been observed when the client is artificially
579
slowed by scheduler constraints.
584
> Ensure that LDAP lookup attempts in <SPAN
588
intended (Laurenz Albe)
596
> to do the right thing when an array
600
> is the target for a FETCH statement returning more
601
than one row, as well as some other array-handling fixes
610
>'s processing of old-style large object
614
> A direct-to-database restore from an archive file generated by a
615
pre-9.0 version of <SPAN
618
> would usually fail if the
619
archive contained more than a few comments for large objects.
627
> for cases where the new server creates
628
a TOAST table but the old version did not (Bruce Momjian)
631
> This rare situation would manifest as <SPAN
633
>"relation OID mismatch"</SPAN
642
>contrib/auto_explain</TT
643
> from changing the output of
653
> is active, it could cause
656
>EXPLAIN (ANALYZE, TIMING OFF)</TT
657
> command to nonetheless
658
print timing information.
663
> Fix query-lifespan memory leak in <TT
674
>contrib/pgcrypto</TT
675
> functions, ensure sensitive
676
information is cleared from stack variables before returning
682
> Prevent use of already-freed memory in
685
>contrib/pgstattuple</TT
697
>contrib/uuid-ossp</TT
698
>, cache the state of the OSSP UUID
699
library across calls (Tom Lane)
702
> This improves the efficiency of UUID generation and reduces the amount
703
of entropy drawn from <TT
712
> Update time zone data files to <SPAN
716
for DST law changes in Crimea, Egypt, and Morocco.
727
SUMMARY="Footer navigation table"
738
HREF="release-9-3.html"
756
HREF="release-9-2-8.html"
b'\\ No newline at end of file'