<!-- doc/src/sgml/release-8.4.sgml -->

<!-- See header comment in release.sgml about typical markup -->

<sect1 id="release-8-4-20">

<title>Release 8.4.20</title>

<note>

<title>Release Date</title>

<simpara>2014-02-20</simpara>

</note>

<para>

This release contains a variety of fixes from 8.4.19.

For information about new features in the 8.4 major release, see

<xref linkend="release-8-4">.

</para>

<para>

The <productname>PostgreSQL</> community will stop releasing updates

for the 8.4.X release series in July 2014.

Users are encouraged to update to a newer release branch soon.

</para>

<sect2>

<title>Migration to Version 8.4.20</title>

<para>

A dump/restore is not required for those running 8.4.X.

</para>

<para>

However, if you are upgrading from a version earlier than 8.4.19,

see <xref linkend="release-8-4-19">.

</para>

</sect2>

<sect2>

<title>Changes</title>

<itemizedlist>

<listitem>

<para>

Fix possible mis-replay of WAL records when some segments of a

relation aren't full size (Greg Stark, Tom Lane)

</para>

<para>

The WAL update could be applied to the wrong page, potentially many

pages past where it should have been. Aside from corrupting data,

this error has been observed to result in significant <quote>bloat</>

of standby servers compared to their masters, due to updates being

applied far beyond where the end-of-file should have been. This

failure mode does not appear to be a significant risk during crash

recovery, only when initially synchronizing a standby created from a

base backup taken from a quickly-changing master.

</para>

</listitem>

<listitem>

<para>

Ensure that insertions into non-leaf GIN index pages write a full-page

WAL record when appropriate (Heikki Linnakangas)

</para>

<para>

The previous coding risked index corruption in the event of a

partial-page write during a system crash.

</para>

</listitem>

<listitem>

<para>

Fix race conditions during server process exit (Robert Haas)

</para>

<para>

Ensure that signal handlers don't attempt to use the

process's <varname>MyProc</> pointer after it's no longer valid.

</para>

</listitem>

<listitem>

<para>

Fix unsafe references to <varname>errno</> within error reporting

logic (Christian Kruse)

</para>

<para>

This would typically lead to odd behaviors such as missing or

inappropriate <literal>HINT</> fields.

</para>

</listitem>

<listitem>

<para>

Fix possible crashes from using <function>ereport()</> too early

during server startup (Tom Lane)

</para>

<para>

The principal case we've seen in the field is a crash if the server

is started in a directory it doesn't have permission to read.

</para>

</listitem>

<listitem>

<para>

Clear retry flags properly in OpenSSL socket write

function (Alexander Kukushkin)

</para>

<para>

This omission could result in a server lockup after unexpected loss

of an SSL-encrypted connection.

</para>

</listitem>

<listitem>

<para>

Fix length checking for Unicode identifiers (<literal>U&amp;"..."</>

syntax) containing escapes (Tom Lane)

</para>

<para>

A spurious truncation warning would be printed for such identifiers

if the escaped form of the identifier was too long, but the

identifier actually didn't need truncation after de-escaping.

</para>

</listitem>

<listitem>

<para>

Fix possible crash due to invalid plan for nested sub-selects, such

as <literal>WHERE (... x IN (SELECT ...) ...) IN (SELECT ...)</>

(Tom Lane)

</para>

</listitem>

<listitem>

<para>

Ensure that <command>ANALYZE</> creates statistics for a table column

even when all the values in it are <quote>too wide</> (Tom Lane)

</para>

<para>

<command>ANALYZE</> intentionally omits very wide values from its

histogram and most-common-values calculations, but it neglected to do

something sane in the case that all the sampled entries are too wide.

</para>

</listitem>

<listitem>

<para>

In <literal>ALTER TABLE ... SET TABLESPACE</>, allow the database's

default tablespace to be used without a permissions check

(Stephen Frost)

</para>

<para>

<literal>CREATE TABLE</> has always allowed such usage,

but <literal>ALTER TABLE</> didn't get the memo.

</para>

</listitem>

<listitem>

<para>

Fix <quote>cannot accept a set</> error when some arms of

a <literal>CASE</> return a set and others don't (Tom Lane)

</para>

</listitem>

<listitem>

<para>

Fix checks for all-zero client addresses in pgstat functions (Kevin

Grittner)

</para>

</listitem>

<listitem>

<para>

Fix possible misclassification of multibyte characters by the text

search parser (Tom Lane)

</para>

<para>

Non-ASCII characters could be misclassified when using C locale with

a multibyte encoding. On Cygwin, non-C locales could fail as well.

</para>

</listitem>

<listitem>

<para>

Fix possible misbehavior in <function>plainto_tsquery()</>

(Heikki Linnakangas)

</para>

<para>

Use <function>memmove()</> not <function>memcpy()</> for copying

overlapping memory regions. There have been no field reports of

this actually causing trouble, but it's certainly risky.

</para>

</listitem>

<listitem>

<para>

Accept <literal>SHIFT_JIS</> as an encoding name for locale checking

purposes (Tatsuo Ishii)

</para>

</listitem>

<listitem>

<para>

Fix misbehavior of <function>PQhost()</> on Windows (Fujii Masao)

</para>

<para>

It should return <literal>localhost</> if no host has been specified.

</para>

</listitem>

<listitem>

<para>

Improve error handling in <application>libpq</> and <application>psql</>

for failures during <literal>COPY TO STDOUT/FROM STDIN</> (Tom Lane)

</para>

<para>

In particular this fixes an infinite loop that could occur in 9.2 and

up if the server connection was lost during <literal>COPY FROM

STDIN</>. Variants of that scenario might be possible in older

versions, or with other client applications.

</para>

</listitem>

<listitem>

<para>

Fix misaligned descriptors in <application>ecpg</> (MauMau)

</para>

</listitem>

<listitem>

<para>

In <application>ecpg</>, handle lack of a hostname in the connection

parameters properly (Michael Meskes)

</para>

</listitem>

<listitem>

<para>

Fix performance regression in <filename>contrib/dblink</> connection

startup (Joe Conway)

</para>

<para>

Avoid an unnecessary round trip when client and server encodings match.

</para>

</listitem>

<listitem>

<para>

In <filename>contrib/isn</>, fix incorrect calculation of the check

digit for ISMN values (Fabien Coelho)

</para>

</listitem>

<listitem>

<para>

Ensure client-code-only installation procedure works as documented

(Peter Eisentraut)

</para>

</listitem>

<listitem>

<para>

In Mingw and Cygwin builds, install the <application>libpq</> DLL

in the <filename>bin</> directory (Andrew Dunstan)

</para>

<para>

This duplicates what the MSVC build has long done. It should fix

problems with programs like <application>psql</> failing to start

because they can't find the DLL.

</para>

</listitem>

<listitem>

<para>

Don't generate plain-text <filename>HISTORY</>

and <filename>src/test/regress/README</> files anymore (Tom Lane)

</para>

<para>

These text files duplicated the main HTML and PDF documentation

formats. The trouble involved in maintaining them greatly outweighs

the likely audience for plain-text format. Distribution tarballs

will still contain files by these names, but they'll just be stubs

directing the reader to consult the main documentation.

The plain-text <filename>INSTALL</> file will still be maintained, as

there is arguably a use-case for that.

</para>

</listitem>

<listitem>

<para>

Update time zone data files to <application>tzdata</> release 2013i

for DST law changes in Jordan and historical changes in Cuba.

</para>

<para>

In addition, the zones <literal>Asia/Riyadh87</>,

<literal>Asia/Riyadh88</>, and <literal>Asia/Riyadh89</> have been

removed, as they are no longer maintained by IANA, and never

represented actual civil timekeeping practice.

</para>

</listitem>

</itemizedlist>

</sect2>

</sect1>

<sect1 id="release-8-4-19">

<title>Release 8.4.19</title>