<!--
-$PostgreSQL: pgsql/doc/src/sgml/release.sgml,v 1.320 2005/01/15 07:53:04 tgl Exp $
+$PostgreSQL: pgsql/doc/src/sgml/release.sgml,v 1.321 2005/01/15 21:11:46 tgl Exp $
-->
<appendix id="release">
<listitem>
<para>
This is the first <productname>PostgreSQL</productname> release
- to natively run on <productname>Microsoft Windows</> as a
- server. It can run as a <productname>Windows</> service. This
+ to run natively on <trademark class=registered>Microsoft Windows</> as
+ a server. It can run as a <productname>Windows</> service. This
release supports NT-based Windows releases like
<productname>Windows 2000</>, <productname>Windows XP</>, and
<productname>Windows 2003</>. Older releases like
systems do not have the infrastructure to support
<productname>PostgreSQL</productname>. A separate installer
project has been created to ease installation on
- <productname>Windows</>: <ulink
+ <productname>Windows</> — see <ulink
url="http://pgfoundry.org/projects/pginstaller">
http://pgfoundry.org/projects/pginstaller</ulink>.
</para>
Although tested throughout our release cycle, the Windows port
does not have the benefit of years of use in production
environments that <productname>PostgreSQL</productname> has on
- Unix platforms and therefore should be treated with the same
+ Unix platforms. Therefore it should be treated with the same
level of caution as you would a new product.
</para>
<listitem>
<para>
- Though <productname>PostgreSQL</productname> is very reliable,
- in previous releases there was no way to recover from disk
+ In previous releases there was no way to recover from disk
drive failure except to restore from a previous backup or use
a standby replication server. Point-in-time recovery allows
continuous backup of the server. You can recover either to
<listitem>
<para>
- Tablespaces allow administrators to select the file systems
- used for storage of tables, indexes, and entire databases.
+ Tablespaces allow administrators to select different file systems
+ for storage of individual tables, indexes, and databases.
This improves performance and control over disk space
usage. Prior releases used <application>initlocation</> and
manual symlink management for such tasks.
<listitem>
<para>
+ In <option>READ COMMITTED</> serialization mode, volatile functions
+ now see the results of concurrent transactions committed up to the
+ beginning of each statement within the function, rather than up to the
+ beginning of the interactive command that called the function.
+ </para>
+ </listitem>
+
+ <listitem>
+ <para>
+ Functions declared <option>STABLE</> or <option>IMMUTABLE</> always
+ use the snapshot of the calling query, and therefore do not see the
+ effects of actions taken after the calling query starts, whether in
+ their own transaction or other transactions. Such a function must be
+ read-only, too, meaning that it cannot use any SQL commands other than
+ <command>SELECT</>.
+ </para>
+ </listitem>
+
+ <listitem>
+ <para>
+ Non-deferred <option>AFTER</> triggers are now fired immediately
+ after completion of the triggering query, rather than upon
+ finishing the current interactive command. This makes a
+ difference when the triggering query occurred within a function:
+ the trigger is invoked before the function proceeds to its next
+ operation.
+ </para>
+ </listitem>
+
+ <listitem>
+ <para>
Server configuration parameters <varname>virtual_host</> and
<varname>tcpip_socket</> have been replaced with a more general
parameter <varname>listen_addresses</>. Also, the server now listens on
<para>
Server configuration parameters <varname>log_pid</>,
<varname>log_timestamp</>, and <varname>log_source_port</> have been
- removed now that a more flexible <varname>log_line_prefix</> has been
- added.
+ replaced with a more general parameter <varname>log_line_prefix</>.
</para>
</listitem>
<listitem>
<para>
- Server configuration parameter <varname>syslog</> has been removed and
+ Server configuration parameter <varname>syslog</> has been
replaced with a more logical <varname>log_destination</> variable to
control the log output destination.
</para>
<para>
Server configuration parameter <varname>log_statement</> has been
changed so it can selectively log just database modification or
- data definition statements. Server configuration parameter <varname>
- log_duration</> now prints only when <varname>log_statement</>
+ data definition statements. Server configuration parameter
+ <varname>log_duration</> now prints only when <varname>log_statement</>
prints the query.
</para>
</listitem>
<listitem>
<para>
- Overflow in integer arithmetic operations is now detected and
- reported as an error.
+ Syntax checking of array input values has been tightened up
+ considerably. Junk that was previously allowed in odd places with
+ odd results now causes an error. Empty-string element values
+ must now be written as <literal>""</>, rather than writing nothing.
+ Also changed behavior with respect to whitespace surrounding
+ array elements: trailing whitespace is now ignored, for symmetry
+ with leading whitespace (which has always been ignored).
</para>
</listitem>
<listitem>
<para>
- The arithmetic operators associated with the single-byte
- <type>"char"</> data type have been removed.
+ Overflow in integer arithmetic operations is now detected and
+ reported as an error.
</para>
</listitem>
<listitem>
<para>
- The server now warns of empty strings passed to
- <type>oid</type>/<type>float4</type>/<type>float8</type> data
- types. In the next major release, doing this will generate an
- error.
+ The arithmetic operators associated with the single-byte
+ <type>"char"</> data type have been removed.
</para>
</listitem>
</para>
</listitem>
+ <listitem>
+ <para>
+ <type>CIDR</> values now must have their non-masked bits be zero.
+ For example, we no longer allow
+ <literal>204.248.199.1/31</literal> as a <type>CIDR</> value. Such
+ values should never have been accepted by
+ <productname>PostgreSQL</productname> and will now be rejected.
+ </para>
+ </listitem>
+
+ <listitem>
+ <para>
+ <command>EXECUTE</command> now returns a completion tag that
+ matches the executed statement.
+ </para>
+ </listitem>
+
<listitem>
<para>
<application>psql</>'s <command>\copy</> command now reads or
</listitem>
<listitem>
- <para> The JDBC client interface has been removed from the core
+ <para>
+ The JDBC client interface has been removed from the core
distribution, and is now hosted at <ulink url=
"http://jdbc.postgresql.org">http://jdbc.postgresql.org</ulink>.
+ </para>
+ </listitem>
+
+ <listitem>
+ <para>
The Tcl client interface has also been removed. There are several
Tcl interfaces now hosted at <ulink url=
"http://gborg.postgresql.org">http://gborg.postgresql.org</ulink>.
one supplied by the operating system. This will provide consistent
behavior across all platforms. In most cases, there should be
little noticeable difference in time zone behavior, except that
- the time zone names used by SET/SHOW <varname>TimeZone</> may
+ the time zone names used by <command>SET</>/<command>SHOW</>
+ <varname>TimeZone</> may
be different from what your platform provides.
</para>
</listitem>
<listitem>
<para>
- <command>EXECUTE</command> now returns a completion tag that
- matches the executed statement.
- </para>
- </listitem>
-
- <listitem>
- <para>
<application>Configure</>'s threading option no longer requires
users to run tests or edit configuration files; threading options
are now detected automatically.
</para>
</listitem>
- <listitem>
- <para>
- Syntax checking of array input values has been tightened up
- considerably. Junk that was previously allowed in odd places with
- odd results now causes an error. Empty-string element values
- must now be written as <literal>""</>, rather than writing nothing.
- Also changed behavior with respect to whitespace surrounding
- array elements: trailing whitespace is now ignored, for symmetry
- with leading whitespace (which has always been ignored).
- </para>
- </listitem>
-
- <listitem>
- <para>
- In <option>READ COMMITTED</> serialization mode, volatile functions
- now see the results of concurrent transactions committed up to the
- beginning of each statement within the function, rather than up to the
- beginning of the interactive command that called the function.
- </para>
- </listitem>
-
- <listitem>
- <para>
- Functions declared <option>STABLE</> or <option>IMMUTABLE</> always
- use the snapshot of the calling query, and therefore do not see the
- effects of actions taken after the calling query starts, whether in
- their own transaction or other transactions. Such a function must be
- read-only, too, meaning that it cannot use any SQL commands other than
- <command>SELECT</>.
- </para>
- </listitem>
-
- <listitem>
- <para>
- Non-deferred <option>AFTER</> triggers are now fired immediately
- after completion of the triggering query, rather than upon
- finishing the current interactive command. This makes a
- difference when the triggering query occurred within a function:
- the trigger is invoked before the function proceeds to its next
- operation.
- </para>
- </listitem>
-
- <listitem>
- <para>
- <type>CIDR</> values now must have their non-masked bits be zero.
- For example, we no longer allow
- <literal>204.248.199.1/31</literal> as a <type>CIDR</> value. Such
- values should never have been accepted by
- <productname>PostgreSQL</productname> and will now be rejected.
- </para>
- </listitem>
-
-
</itemizedlist>
</para>
</sect2>
<listitem>
<para>
The 8.1 release will remove the function
- <function>to_char(interval)</>.
+ <function>to_char(interval, text)</>.
</para>
</listitem>
+ <listitem>
+ <para>
+ The server now warns of empty strings passed to
+ <type>oid</type>/<type>float4</type>/<type>float8</type> data
+ types, but continues to interpret them as zeroes as before.
+ In the next major release, empty strings will be considered
+ invalid input for these data types.
+ </para>
+ </listitem>
+
<listitem>
<para>
By default, tables in <productname>PostgreSQL</productname> 8.0
this will <emphasis>not</emphasis> be the case: to create a table
that contains <type>OID</>s, the <option>WITH OIDS</> clause must
be specified or the <varname>default_with_oids</varname>
- configuration parameter must be enabled. Users are encouraged to
+ configuration parameter must be set. Users are encouraged to
explicitly specify <option>WITH OIDS</> if their tables
require OIDs for compatibility with future releases of
<productname>PostgreSQL</productname>.
In previous releases, the checkpoint process, which runs every few
minutes, would write all dirty buffers to the operating system's
buffer cache then flush all dirty operating system buffers to
- disk. This often resulted in a periodic spike in disk usage that
+ disk. This resulted in a periodic spike in disk usage that often
hurt performance. The new code uses a background writer to trickle
disk writes at a steady pace so checkpoints have far fewer dirty
pages to write to disk. Also, the new code does not issue a global
Use dynamically-generated table size estimates while planning (Tom)
</para>
<para>
- The optimizer now uses a table's current actual size on disk as its
- estimate of the number of blocks in the table, and it makes an estimate
- of the number of rows in the table based on the current size on disk.
- Formerly, the
- <structname>pg_class</structname>.<structfield>relpages</structfield>
- and
- <structname>pg_class</structname>.<structfield>reltuples</structfield>
- fields were used as-is, but these values might be quite out-of-date,
- leading to poor choices of plans. They are now treated only as an
- indication of the table's density (rows per page).
+ Formerly the planner estimated table sizes using the values seen
+ by the last <command>VACUUM</command> or <command>ANALYZE</command>,
+ both as to physical table size (number of pages) and number of rows.
+ Now, the current physical table size is obtained from the kernel,
+ and the number of rows is estimated by multiplying the table size
+ by the row density (rows per page) seen by the last
+ <command>VACUUM</command> or <command>ANALYZE</command>. This should
+ produce more reliable estimates in cases where the table size has
+ changed significantly since the last housekeeping command.
</para>
</listitem>
<listitem>
<para>
- Allow collection of <command>ANALYZE</command> statistics for
+ <command>ANALYZE</command> now collects statistics for
expression indexes (Tom)
</para>
<para>
</para>
</listitem>
- <listitem>
- <para>
- Improve optimizer rowcount estimates (Tom)
- </para>
- <para>
- Formerly the planner estimated table sizes using the values seen
- by the last <command>VACUUM</command> or <command>ANALYZE</command>,
- both as to physical table size (number of pages) and number of rows.
- Now, the physical table size is obtained directly from the kernel,
- and the number of rows is estimated by multiplying the table size
- by the row density (rows per page) seen by the last
- <command>VACUUM</command> or <command>ANALYZE</command>. This should
- produce more reliable estimates in cases where the table size has
- changed significantly since the last housekeeping command.
- </para>
- </listitem>
-
</itemizedlist>
</sect3>
<listitem>
<para>
- Add new read-only server configuration parameters to query server
+ Add new read-only server configuration parameters to show server
compile-time settings: <varname>block_size</>,
<varname>integer_datetimes</>, <varname>max_function_args</>,
<varname>max_identifier_length</>, <varname>max_index_keys</> (Joe)
parameters with a unified <varname>listen_addresses</> parameter
(Andrew, Tom)
</para>
+ <para>
+ <varname>virtual_host</> could only specify a single IP address to
+ listen on. <varname>listen_addresses</> allows multiple addresses
+ to be specified.
+ </para>
</listitem>
<listitem>
parameters to be used in the query. In this release, planning of
unnamed prepared statements is delayed until the first execution,
and the actual parameter values of that execution are used as
- optimization hints.
+ optimization hints. This allows use of out-of-line parameter passing
+ without incurring a performance penalty.
</para>
</listitem>
control whether tables are created with <type>OID</>s by default (Neil)
</para>
<para>
- This allows administrators to default all <command>CREATE
- TABLE</command> commands to create tables without <type>OID</>
- columns.
+ This allows administrators to control whether <command>CREATE
+ TABLE</command> commands create tables with or without <type>OID</>
+ columns by default. (Note: the current factory default setting for
+ <varname>default_with_oids</> is <literal>TRUE</>, but the default
+ will become <literal>FALSE</> in future releases.)
</para>
</listitem>
<listitem>
<para>
- Constraint/Index/<type>SERIAL</> names are now table_column_type
+ Constraint/Index/<type>SERIAL</> names are now
+ <replaceable>table_column_type</>
with numbers appended to guarantee uniqueness within the schema
(Tom)
</para>
</para>
<para>
<productname>PostgreSQL</productname> uses the user name as salt
- when encrypting passwords via MD5. When a user name is changed,
- their salt no longer matches the stored MD5 password, so the
+ when encrypting passwords via MD5. When a user's name is changed,
+ the salt will no longer match the stored MD5 password, so the
stored password becomes useless. In this release a notice is
generated and the password is cleared. A new password must then
- be assigned.
+ be assigned if the user is to be able to log in with a password.
</para>
</listitem>
Formerly, the parser would select these operators in many situations
where an <quote>unable to select an operator</> error would be more
appropriate, such as <literal>null * null</>. If you actually want
- to do arithmetic on a <type>"char"</> column, you can cast it to integer.
+ to do arithmetic on a <type>"char"</> column, you can cast it to
+ integer explicitly.
</para>
</listitem>
<listitem>
<para>
- Warn of empty string being passed to
+ Warn about empty string being passed to
<type>OID</>/<type>float4</>/<type>float8</> data types (Neil)
</para>
<para>
<listitem>
<para>
- Allow
- <type>int2</>/<type>int4</>/<type>int8</>/<type>float4</>/<type>float8</>
- input routines to have leading or trailing whitespace
+ Allow leading or trailing whitespace in
+ <type>int2</>/<type>int4</>/<type>int8</>/<type>float4</>/<type>float8</>
+ input routines
(Neil)
</para>
</listitem>
effects of actions taken after the calling query starts, whether in
their own transaction or other transactions. Such a function must be
read-only, too, meaning that it cannot use any SQL commands other than
- <command>SELECT</>.
+ <command>SELECT</>. There is a considerable performance gain from
+ declaring a function <literal>STABLE</> or <literal>IMMUTABLE</>
+ rather than <literal>VOLATILE</>.
</para>
</listitem>
<listitem>
<para>
- Make libpq <option>SIGPIPE</> thread-safe (Bruce)
+ Make libpq's <option>SIGPIPE</> handling thread-safe (Bruce)
</para>
</listitem>
<listitem>
<para>
- Removed <filename>contrib/pg_logger</>
+ Removed <filename>contrib/pg_logger</>: obsoleted by integrated logging
+ subprocess
</para>
</listitem>