| Current File : //usr/share/doc/postgresql-9.2.24/html/release-8-1.html |
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<HTML
><HEAD
><TITLE
>Release 8.1</TITLE
><META
NAME="GENERATOR"
CONTENT="Modular DocBook HTML Stylesheet Version 1.79"><LINK
REV="MADE"
HREF="mailto:pgsql-docs@postgresql.org"><LINK
REL="HOME"
TITLE="PostgreSQL 9.2.24 Documentation"
HREF="index.html"><LINK
REL="UP"
TITLE="Release Notes"
HREF="release.html"><LINK
REL="PREVIOUS"
TITLE="Release 8.1.1"
HREF="release-8-1-1.html"><LINK
REL="NEXT"
TITLE="Release 8.0.26"
HREF="release-8-0-26.html"><LINK
REL="STYLESHEET"
TYPE="text/css"
HREF="stylesheet.css"><META
HTTP-EQUIV="Content-Type"
CONTENT="text/html; charset=ISO-8859-1"><META
NAME="creation"
CONTENT="2017-11-06T22:43:11"></HEAD
><BODY
CLASS="SECT1"
><DIV
CLASS="NAVHEADER"
><TABLE
SUMMARY="Header navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TH
COLSPAN="5"
ALIGN="center"
VALIGN="bottom"
><A
HREF="index.html"
>PostgreSQL 9.2.24 Documentation</A
></TH
></TR
><TR
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
TITLE="Release 8.1.1"
HREF="release-8-1-1.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
HREF="release.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="60%"
ALIGN="center"
VALIGN="bottom"
>Appendix E. Release Notes</TD
><TD
WIDTH="20%"
ALIGN="right"
VALIGN="top"
><A
TITLE="Release 8.0.26"
HREF="release-8-0-26.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
></TABLE
><HR
ALIGN="LEFT"
WIDTH="100%"></DIV
><DIV
CLASS="SECT1"
><H1
CLASS="SECT1"
><A
NAME="RELEASE-8-1"
>E.169. Release 8.1</A
></H1
><DIV
CLASS="FORMALPARA"
><P
><B
>Release date: </B
>2005-11-08</P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN136942"
>E.169.1. Overview</A
></H2
><P
> Major changes in this release:
</P
><P
></P
><DIV
CLASS="VARIABLELIST"
><DL
><DT
>Improve concurrent access to the shared buffer cache (Tom)</DT
><DD
><P
> Access to the shared buffer cache was identified as a
significant scalability problem, particularly on multi-CPU
systems. In this release, the way that locking is done in the
buffer manager has been overhauled to reduce lock contention
and improve scalability. The buffer manager has also been
changed to use a <SPAN
CLASS="QUOTE"
>"clock sweep"</SPAN
> replacement
policy.
</P
></DD
><DT
>Allow index scans to use an intermediate in-memory bitmap (Tom)</DT
><DD
><P
> In previous releases, only a single index could be used to do
lookups on a table. With this feature, if a query has
<TT
CLASS="COMMAND"
>WHERE tab.col1 = 4 and tab.col2 = 9</TT
>, and there is
no multicolumn index on <TT
CLASS="LITERAL"
>col1</TT
> and <TT
CLASS="LITERAL"
>col2</TT
>,
but there is an index on <TT
CLASS="LITERAL"
>col1</TT
> and another on
<TT
CLASS="LITERAL"
>col2</TT
>, it is possible to search both indexes and
combine the results in memory, then do heap fetches for only
the rows matching both the <TT
CLASS="LITERAL"
>col1</TT
> and
<TT
CLASS="LITERAL"
>col2</TT
> restrictions. This is very useful in
environments that have a lot of unstructured queries where it
is impossible to create indexes that match all possible access
conditions. Bitmap scans are useful even with a single index,
as they reduce the amount of random access needed; a bitmap
index scan is efficient for retrieving fairly large fractions
of the complete table, whereas plain index scans are not.
</P
></DD
><DT
>Add two-phase commit (Heikki Linnakangas, Alvaro, Tom)</DT
><DD
><P
> Two-phase commit allows transactions to be "prepared" on several
computers, and once all computers have successfully prepared
their transactions (none failed), all transactions can be
committed. Even if a machine crashes after a prepare, the
prepared transaction can be committed after the machine is
restarted. New syntax includes <TT
CLASS="COMMAND"
>PREPARE TRANSACTION</TT
> and
<TT
CLASS="COMMAND"
>COMMIT/ROLLBACK PREPARED</TT
>. A new system view
<TT
CLASS="LITERAL"
>pg_prepared_xacts</TT
> has also been added.
</P
></DD
><DT
>Create a new role system that replaces users and groups
(Stephen Frost)</DT
><DD
><P
> Roles are a combination of users and groups. Like users, they
can have login capability, and like groups, a role can have
other roles as members. Roles basically remove the distinction
between users and groups. For example, a role can:
</P
><P
></P
><UL
><LI
><P
> Have login capability (optionally)
</P
></LI
><LI
><P
> Own objects
</P
></LI
><LI
><P
> Hold access permissions for database objects
</P
></LI
><LI
><P
> Inherit permissions from other roles it is a member of
</P
></LI
></UL
><P
> Once a user logs into a role, she obtains capabilities of
the login role plus any inherited roles, and can use
<TT
CLASS="COMMAND"
>SET ROLE</TT
> to switch to other roles she is a member of.
This feature is a generalization of the SQL standard's concept of
roles.
This change also replaces <TT
CLASS="STRUCTNAME"
>pg_shadow</TT
> and
<TT
CLASS="STRUCTNAME"
>pg_group</TT
> by new role-capable catalogs
<TT
CLASS="STRUCTNAME"
>pg_authid</TT
> and <TT
CLASS="STRUCTNAME"
>pg_auth_members</TT
>. The old
tables are redefined as read-only views on the new role tables.
</P
></DD
><DT
>Automatically use indexes for <CODE
CLASS="FUNCTION"
>MIN()</CODE
> and
<CODE
CLASS="FUNCTION"
>MAX()</CODE
> (Tom)</DT
><DD
><P
> In previous releases, the only way to use an index for
<CODE
CLASS="FUNCTION"
>MIN()</CODE
> or <CODE
CLASS="FUNCTION"
>MAX()</CODE
> was to rewrite the
query as <TT
CLASS="COMMAND"
>SELECT col FROM tab ORDER BY col LIMIT 1</TT
>.
Index usage now happens automatically.
</P
></DD
><DT
>Move <TT
CLASS="FILENAME"
>/contrib/pg_autovacuum</TT
> into the main server
(Alvaro)</DT
><DD
><P
> Integrating autovacuum into the server allows it to be
automatically started and stopped in sync with the database
server, and allows autovacuum to be configured from
<TT
CLASS="FILENAME"
>postgresql.conf</TT
>.
</P
></DD
><DT
>Add shared row level locks using <TT
CLASS="COMMAND"
>SELECT ... FOR SHARE</TT
>
(Alvaro)</DT
><DD
><P
> While <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
>'s MVCC locking
allows <TT
CLASS="COMMAND"
>SELECT</TT
> to never be blocked by writers and
therefore does not need shared row locks for typical operations,
shared locks are useful for applications that require shared row
locking. In particular this reduces the locking requirements
imposed by referential integrity checks.
</P
></DD
><DT
>Add dependencies on shared objects, specifically roles
(Alvaro)</DT
><DD
><P
> This extension of the dependency mechanism prevents roles from
being dropped while there are still database objects they own.
Formerly it was possible to accidentally <SPAN
CLASS="QUOTE"
>"orphan"</SPAN
> objects by
deleting their owner. While this could be recovered from, it
was messy and unpleasant.
</P
></DD
><DT
>Improve performance for partitioned tables (Simon)</DT
><DD
><P
> The new <TT
CLASS="VARNAME"
>constraint_exclusion</TT
> configuration
parameter avoids lookups on child tables where constraints indicate
that no matching rows exist in the child table.
</P
><P
> This allows for a basic type of table partitioning. If child tables
store separate key ranges and this is enforced using appropriate
<TT
CLASS="COMMAND"
>CHECK</TT
> constraints, the optimizer will skip child
table accesses when the constraint guarantees no matching rows
exist in the child table.
</P
></DD
></DL
></DIV
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN137022"
>E.169.2. Migration to Version 8.1</A
></H2
><P
> A dump/restore using <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> is required
for those wishing to migrate data from any previous release.
</P
><P
> The 8.0 release announced that the <CODE
CLASS="FUNCTION"
>to_char()</CODE
> function
for intervals would be removed in 8.1. However, since no better API
has been suggested, <CODE
CLASS="FUNCTION"
>to_char(interval)</CODE
> has been enhanced in
8.1 and will remain in the server.
</P
><P
> Observe the following incompatibilities:
</P
><P
></P
><UL
><LI
><P
> <TT
CLASS="VARNAME"
>add_missing_from</TT
> is now false by default (Neil)
</P
><P
> By default, we now generate an error if a table is used in a query
without a <TT
CLASS="LITERAL"
>FROM</TT
> reference. The old behavior is still
available, but the parameter must be set to 'true' to obtain it.
</P
><P
> It might be necessary to set <TT
CLASS="VARNAME"
>add_missing_from</TT
> to true
in order to load an existing dump file, if the dump contains any
views or rules created using the implicit-<TT
CLASS="LITERAL"
>FROM</TT
> syntax.
This should be a one-time annoyance, because
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> 8.1 will convert
such views and rules to standard explicit-<TT
CLASS="LITERAL"
>FROM</TT
> syntax.
Subsequent dumps will therefore not have the problem.
</P
></LI
><LI
><P
> Cause input of a zero-length string (<TT
CLASS="LITERAL"
>''</TT
>) for
<TT
CLASS="TYPE"
>float4</TT
>/<TT
CLASS="TYPE"
>float8</TT
>/<TT
CLASS="TYPE"
>oid</TT
>
to throw an error, rather than treating it as a zero (Neil)
</P
><P
> This change is consistent with the current handling of
zero-length strings for integers. The schedule for this change
was announced in 8.0.
</P
></LI
><LI
><P
> <TT
CLASS="VARNAME"
>default_with_oids</TT
> is now false by default (Neil)
</P
><P
> With this option set to false, user-created tables no longer
have an OID column unless <TT
CLASS="COMMAND"
>WITH OIDS</TT
> is specified in
<TT
CLASS="COMMAND"
>CREATE TABLE</TT
>. Though OIDs have existed in all
releases of <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
>, their use is limited
because they are only four bytes long and the counter is shared
across all installed databases. The preferred way of uniquely
identifying rows is via sequences and the <TT
CLASS="TYPE"
>SERIAL</TT
> type,
which have been supported since <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> 6.4.
</P
></LI
><LI
><P
> Add <TT
CLASS="LITERAL"
>E''</TT
> syntax so eventually ordinary strings can
treat backslashes literally (Bruce)
</P
><P
> Currently <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> processes a
backslash in a string literal as introducing a special escape sequence,
e.g. <TT
CLASS="LITERAL"
>\n</TT
> or <TT
CLASS="LITERAL"
>\010</TT
>.
While this allows easy entry of special values, it is
nonstandard and makes porting of applications from other
databases more difficult. For this reason, the
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> project is planning to
remove the special meaning of backslashes in strings. For
backward compatibility and for users who want special backslash
processing, a new string syntax has been created. This new string
syntax is formed by writing an <TT
CLASS="LITERAL"
>E</TT
> immediately preceding the
single quote that starts the string, e.g. <TT
CLASS="LITERAL"
>E'hi\n'</TT
>. While
this release does not change the handling of backslashes in strings, it
does add new configuration parameters to help users migrate applications
for future releases:
</P
><P
></P
><UL
><LI
><P
> <TT
CLASS="VARNAME"
>standard_conforming_strings</TT
> — does this release
treat backslashes literally in ordinary strings?
</P
></LI
><LI
><P
> <TT
CLASS="VARNAME"
>escape_string_warning</TT
> — warn about backslashes in
ordinary (non-E) strings
</P
></LI
></UL
><P
> The <TT
CLASS="VARNAME"
>standard_conforming_strings</TT
> value is read-only.
Applications can retrieve the value to know how backslashes are
processed. (Presence of the parameter can also be taken as an
indication that <TT
CLASS="LITERAL"
>E''</TT
> string syntax is supported.)
In a future release, <TT
CLASS="VARNAME"
>standard_conforming_strings</TT
>
will be true, meaning backslashes will be treated literally in
non-E strings. To prepare for this change, use <TT
CLASS="LITERAL"
>E''</TT
>
strings in places that need special backslash processing, and
turn on <TT
CLASS="VARNAME"
>escape_string_warning</TT
> to find additional
strings that need to be converted to use <TT
CLASS="LITERAL"
>E''</TT
>.
Also, use two single-quotes (<TT
CLASS="LITERAL"
>''</TT
>) to embed a literal
single-quote in a string, rather than the
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
>-supported syntax of
backslash single-quote (<TT
CLASS="LITERAL"
>\'</TT
>). The former is
standards-conforming and does not require the use of the
<TT
CLASS="LITERAL"
>E''</TT
> string syntax. You can also use the
<TT
CLASS="LITERAL"
>$$</TT
> string syntax, which does not treat backslashes
specially.
</P
></LI
><LI
><P
> Make <TT
CLASS="COMMAND"
>REINDEX DATABASE</TT
> reindex all indexes in the
database (Tom)
</P
><P
> Formerly, <TT
CLASS="COMMAND"
>REINDEX DATABASE</TT
> reindexed only
system tables. This new behavior seems more intuitive. A new
command <TT
CLASS="COMMAND"
>REINDEX SYSTEM</TT
> provides the old functionality
of reindexing just the system tables.
</P
></LI
><LI
><P
> Read-only large object descriptors now obey MVCC snapshot semantics
</P
><P
> When a large object is opened with <TT
CLASS="LITERAL"
>INV_READ</TT
> (and not
<TT
CLASS="LITERAL"
>INV_WRITE</TT
>), the data read from the descriptor will now
reflect a <SPAN
CLASS="QUOTE"
>"snapshot"</SPAN
> of the large object's state at the
time of the transaction snapshot in use by the query that called
<CODE
CLASS="FUNCTION"
>lo_open()</CODE
>. To obtain the old behavior of always
returning the latest committed data, include <TT
CLASS="LITERAL"
>INV_WRITE</TT
>
in the mode flags for <CODE
CLASS="FUNCTION"
>lo_open()</CODE
>.
</P
></LI
><LI
><P
> Add proper dependencies for arguments of sequence functions (Tom)
</P
><P
> In previous releases, sequence names passed to <CODE
CLASS="FUNCTION"
>nextval()</CODE
>,
<CODE
CLASS="FUNCTION"
>currval()</CODE
>, and <CODE
CLASS="FUNCTION"
>setval()</CODE
> were stored as
simple text strings, meaning that renaming or dropping a
sequence used in a <TT
CLASS="LITERAL"
>DEFAULT</TT
> clause made the clause
invalid. This release stores all newly-created sequence function
arguments as internal OIDs, allowing them to track sequence
renaming, and adding dependency information that prevents
improper sequence removal. It also makes such <TT
CLASS="LITERAL"
>DEFAULT</TT
>
clauses immune to schema renaming and search path changes.
</P
><P
> Some applications might rely on the old behavior of
run-time lookup for sequence names. This can still be done by
explicitly casting the argument to <TT
CLASS="TYPE"
>text</TT
>, for example
<TT
CLASS="LITERAL"
>nextval('myseq'::text)</TT
>.
</P
><P
> Pre-8.1 database dumps loaded into 8.1 will use the old text-based
representation and therefore will not have the features of
OID-stored arguments. However, it is possible to update a
database containing text-based <TT
CLASS="LITERAL"
>DEFAULT</TT
> clauses.
First, save this query into a file, such as <TT
CLASS="FILENAME"
>fixseq.sql</TT
>:
</P><PRE
CLASS="PROGRAMLISTING"
>SELECT 'ALTER TABLE ' ||
pg_catalog.quote_ident(n.nspname) || '.' ||
pg_catalog.quote_ident(c.relname) ||
' ALTER COLUMN ' || pg_catalog.quote_ident(a.attname) ||
' SET DEFAULT ' ||
regexp_replace(d.adsrc,
$$val\(\(('[^']*')::text\)::regclass$$,
$$val(\1$$,
'g') ||
';'
FROM pg_namespace n, pg_class c, pg_attribute a, pg_attrdef d
WHERE n.oid = c.relnamespace AND
c.oid = a.attrelid AND
a.attrelid = d.adrelid AND
a.attnum = d.adnum AND
d.adsrc ~ $$val\(\('[^']*'::text\)::regclass$$;</PRE
><P>
Next, run the query against a database to find what
adjustments are required, like this for database <TT
CLASS="LITERAL"
>db1</TT
>:
</P><PRE
CLASS="PROGRAMLISTING"
>psql -t -f fixseq.sql db1</PRE
><P>
This will show the <TT
CLASS="COMMAND"
>ALTER TABLE</TT
> commands needed to
convert the database to the newer OID-based representation.
If the commands look reasonable, run this to update the database:
</P><PRE
CLASS="PROGRAMLISTING"
>psql -t -f fixseq.sql db1 | psql -e db1</PRE
><P>
This process must be repeated in each database to be updated.
</P
></LI
><LI
><P
> In <SPAN
CLASS="APPLICATION"
>psql</SPAN
>, treat unquoted
<TT
CLASS="LITERAL"
>\{digit}+</TT
> sequences as octal (Bruce)
</P
><P
> In previous releases, <TT
CLASS="LITERAL"
>\{digit}+</TT
> sequences were
treated as decimal, and only <TT
CLASS="LITERAL"
>\0{digit}+</TT
> were treated
as octal. This change was made for consistency.
</P
></LI
><LI
><P
> Remove grammar productions for prefix and postfix <TT
CLASS="LITERAL"
>%</TT
>
and <TT
CLASS="LITERAL"
>^</TT
> operators
(Tom)
</P
><P
> These have never been documented and complicated the use of the
modulus operator (<TT
CLASS="LITERAL"
>%</TT
>) with negative numbers.
</P
></LI
><LI
><P
> Make <TT
CLASS="LITERAL"
>&<</TT
> and <TT
CLASS="LITERAL"
>&></TT
> for polygons
consistent with the box "over" operators (Tom)
</P
></LI
><LI
><P
> <TT
CLASS="COMMAND"
>CREATE LANGUAGE</TT
> can ignore the provided arguments
in favor of information from <TT
CLASS="STRUCTNAME"
>pg_pltemplate</TT
>
(Tom)
</P
><P
> A new system catalog <TT
CLASS="STRUCTNAME"
>pg_pltemplate</TT
> has been defined
to carry information about the preferred definitions of procedural
languages (such as whether they have validator functions). When
an entry exists in this catalog for the language being created,
<TT
CLASS="COMMAND"
>CREATE LANGUAGE</TT
> will ignore all its parameters except the
language name and instead use the catalog information. This measure
was taken because of increasing problems with obsolete language
definitions being loaded by old dump files. As of 8.1,
<SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> will dump procedural language definitions as
just <TT
CLASS="COMMAND"
>CREATE LANGUAGE <TT
CLASS="REPLACEABLE"
><I
>name</I
></TT
></TT
>, relying
on a template entry to exist at load time. We expect this will be a
more future-proof representation.
</P
></LI
><LI
><P
> Make <CODE
CLASS="FUNCTION"
>pg_cancel_backend(int)</CODE
> return a
<TT
CLASS="TYPE"
>boolean</TT
> rather than an <TT
CLASS="TYPE"
>integer</TT
> (Neil)
</P
></LI
><LI
><P
> Some users are having problems loading UTF-8 data into 8.1.X.
This is because previous versions allowed invalid UTF-8 byte
sequences to be entered into the database, and this release
properly accepts only valid UTF-8 sequences. One way to correct a
dumpfile is to run the command <TT
CLASS="COMMAND"
>iconv -c -f UTF-8 -t
UTF-8 -o cleanfile.sql dumpfile.sql</TT
>. The <TT
CLASS="LITERAL"
>-c</TT
> option
removes invalid character sequences. A diff of the two files will
show the sequences that are invalid. <TT
CLASS="COMMAND"
>iconv</TT
> reads the
entire input file into memory so it might be necessary to use
<SPAN
CLASS="APPLICATION"
>split</SPAN
> to break up the dump into multiple smaller
files for processing.
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN137158"
>E.169.3. Additional Changes</A
></H2
><P
> Below you will find a detailed account of the additional changes
between <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> 8.1 and the
previous major release.
</P
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137162"
>E.169.3.1. Performance Improvements</A
></H3
><P
></P
><UL
><LI
><P
> Improve GiST and R-tree index performance (Neil)
</P
></LI
><LI
><P
> Improve the optimizer, including auto-resizing of hash joins
(Tom)
</P
></LI
><LI
><P
> Overhaul internal API in several areas
</P
></LI
><LI
><P
> Change WAL record CRCs from 64-bit to 32-bit (Tom)
</P
><P
> We determined that the extra cost of computing 64-bit CRCs was
significant, and the gain in reliability too marginal to justify it.
</P
></LI
><LI
><P
> Prevent writing large empty gaps in WAL pages (Tom)
</P
></LI
><LI
><P
> Improve spinlock behavior on SMP machines, particularly Opterons (Tom)
</P
></LI
><LI
><P
> Allow nonconsecutive index columns to be used in a multicolumn
index (Tom)
</P
><P
> For example, this allows an index on columns a,b,c to be used in
a query with <TT
CLASS="COMMAND"
>WHERE a = 4 and c = 10</TT
>.
</P
></LI
><LI
><P
> Skip WAL logging for <TT
CLASS="COMMAND"
>CREATE TABLE AS</TT
> /
<TT
CLASS="COMMAND"
>SELECT INTO</TT
> (Simon)
</P
><P
> Since a crash during <TT
CLASS="COMMAND"
>CREATE TABLE AS</TT
> would cause the
table to be dropped during recovery, there is no reason to WAL
log as the table is loaded. (Logging still happens if WAL
archiving is enabled, however.)
</P
></LI
><LI
><P
> Allow concurrent GiST index access (Teodor, Oleg)
</P
></LI
><LI
><P
> Add configuration parameter <TT
CLASS="VARNAME"
>full_page_writes</TT
> to
control writing full pages to WAL (Bruce)
</P
><P
> To prevent partial disk writes from corrupting the database,
<SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> writes a complete copy of
each database disk page to WAL the first time it is modified
after a checkpoint. This option turns off that functionality for more
speed. This is safe to use with battery-backed disk caches where
partial page writes cannot happen.
</P
></LI
><LI
><P
> Use <TT
CLASS="LITERAL"
>O_DIRECT</TT
> if available when using
<TT
CLASS="LITERAL"
>O_SYNC</TT
> for <TT
CLASS="VARNAME"
>wal_sync_method</TT
>
(Itagaki Takahiro)
</P
><P
> <TT
CLASS="LITERAL"
>O_DIRECT</TT
> causes disk writes to bypass the kernel
cache, and for WAL writes, this improves performance.
</P
></LI
><LI
><P
> Improve <TT
CLASS="COMMAND"
>COPY FROM</TT
> performance (Alon Goldshuv)
</P
><P
> This was accomplished by reading <TT
CLASS="COMMAND"
>COPY</TT
> input in
larger chunks, rather than character by character.
</P
></LI
><LI
><P
> Improve the performance of <CODE
CLASS="FUNCTION"
>COUNT()</CODE
>,
<CODE
CLASS="FUNCTION"
>SUM</CODE
>, <CODE
CLASS="FUNCTION"
>AVG()</CODE
>,
<CODE
CLASS="FUNCTION"
>STDDEV()</CODE
>, and
<CODE
CLASS="FUNCTION"
>VARIANCE()</CODE
> (Neil, Tom)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137214"
>E.169.3.2. Server Changes</A
></H3
><P
></P
><UL
><LI
><P
> Prevent problems due to transaction ID (XID) wraparound (Tom)
</P
><P
> The server will now warn when the transaction counter approaches
the wraparound point. If the counter becomes too close to wraparound,
the server will stop accepting queries. This ensures that data is
not lost before needed vacuuming is performed.
</P
></LI
><LI
><P
> Fix problems with object IDs (OIDs) conflicting with existing system
objects after the OID counter has wrapped around (Tom)
</P
></LI
><LI
><P
> Add warning about the need to increase
<TT
CLASS="VARNAME"
>max_fsm_relations</TT
> and <TT
CLASS="VARNAME"
>max_fsm_pages</TT
>
during <TT
CLASS="COMMAND"
>VACUUM</TT
> (Ron Mayer)
</P
></LI
><LI
><P
> Add <TT
CLASS="VARNAME"
>temp_buffers</TT
> configuration parameter to allow
users to determine the size of the local buffer area for
temporary table access (Tom)
</P
></LI
><LI
><P
> Add session start time and client IP address to
<TT
CLASS="LITERAL"
>pg_stat_activity</TT
> (Magnus)
</P
></LI
><LI
><P
> Adjust <TT
CLASS="LITERAL"
>pg_stat</TT
> views for bitmap scans (Tom)
</P
><P
> The meanings of some of the fields have changed slightly.
</P
></LI
><LI
><P
> Enhance <TT
CLASS="LITERAL"
>pg_locks</TT
> view (Tom)
</P
></LI
><LI
><P
> Log queries for client-side <TT
CLASS="COMMAND"
>PREPARE</TT
> and
<TT
CLASS="COMMAND"
>EXECUTE</TT
> (Simon)
</P
></LI
><LI
><P
> Allow Kerberos name and user name case sensitivity to be
specified in <TT
CLASS="FILENAME"
>postgresql.conf</TT
> (Magnus)
</P
></LI
><LI
><P
> Add configuration parameter <TT
CLASS="VARNAME"
>krb_server_hostname</TT
> so
that the server host name can be specified as part of service
principal (Todd Kover)
</P
><P
> If not set, any service principal matching an entry in the
keytab can be used. This is new Kerberos matching behavior in
this release.
</P
></LI
><LI
><P
> Add <TT
CLASS="VARNAME"
>log_line_prefix</TT
> options for millisecond
timestamps (<TT
CLASS="LITERAL"
>%m</TT
>) and remote host (<TT
CLASS="LITERAL"
>%h</TT
>) (Ed
L.)
</P
></LI
><LI
><P
> Add WAL logging for GiST indexes (Teodor, Oleg)
</P
><P
> GiST indexes are now safe for crash and point-in-time recovery.
</P
></LI
><LI
><P
> Remove old <TT
CLASS="FILENAME"
>*.backup</TT
> files when we do
<CODE
CLASS="FUNCTION"
>pg_stop_backup()</CODE
> (Bruce)
</P
><P
> This prevents a large number of <TT
CLASS="FILENAME"
>*.backup</TT
> files from
existing in <TT
CLASS="FILENAME"
>pg_xlog/</TT
>.
</P
></LI
><LI
><P
> Add configuration parameters to control TCP/IP keep-alive
times for idle, interval, and count (Oliver Jowett)
</P
><P
> These values can be changed to allow more rapid detection of
lost client connections.
</P
></LI
><LI
><P
> Add per-user and per-database connection limits (Petr Jelinek)
</P
><P
> Using <TT
CLASS="COMMAND"
>ALTER USER</TT
> and <TT
CLASS="COMMAND"
>ALTER DATABASE</TT
>,
limits can now be enforced on the maximum number of sessions that
can concurrently connect as a specific user or to a specific database.
Setting the limit to zero disables user or database connections.
</P
></LI
><LI
><P
> Allow more than two gigabytes of shared memory and per-backend
work memory on 64-bit machines (Koichi Suzuki)
</P
></LI
><LI
><P
> New system catalog <TT
CLASS="STRUCTNAME"
>pg_pltemplate</TT
> allows overriding
obsolete procedural-language definitions in dump files (Tom)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137279"
>E.169.3.3. Query Changes</A
></H3
><P
></P
><UL
><LI
><P
> Add temporary views (Koju Iijima, Neil)
</P
></LI
><LI
><P
> Fix <TT
CLASS="COMMAND"
>HAVING</TT
> without any aggregate functions or
<TT
CLASS="COMMAND"
>GROUP BY</TT
> so that the query returns a single group (Tom)
</P
><P
> Previously, such a case would treat the <TT
CLASS="COMMAND"
>HAVING</TT
>
clause the same as a <TT
CLASS="COMMAND"
>WHERE</TT
> clause. This was not per spec.
</P
></LI
><LI
><P
> Add <TT
CLASS="COMMAND"
>USING</TT
> clause to allow additional tables to be
specified to <TT
CLASS="COMMAND"
>DELETE</TT
> (Euler Taveira de Oliveira, Neil)
</P
><P
> In prior releases, there was no clear method for specifying
additional tables to be used for joins in a <TT
CLASS="COMMAND"
>DELETE</TT
>
statement. <TT
CLASS="COMMAND"
>UPDATE</TT
> already has a <TT
CLASS="LITERAL"
>FROM</TT
>
clause for this purpose.
</P
></LI
><LI
><P
> Add support for <TT
CLASS="LITERAL"
>\x</TT
> hex escapes in backend and ecpg
strings (Bruce)
</P
><P
> This is just like the standard C <TT
CLASS="LITERAL"
>\x</TT
> escape syntax.
Octal escapes were already supported.
</P
></LI
><LI
><P
> Add <TT
CLASS="COMMAND"
>BETWEEN SYMMETRIC</TT
> query syntax (Pavel Stehule)
</P
><P
> This feature allows <TT
CLASS="COMMAND"
>BETWEEN</TT
> comparisons without
requiring the first value to be less than the second. For
example, <TT
CLASS="COMMAND"
>2 BETWEEN [ASYMMETRIC] 3 AND 1</TT
> returns
false, while <TT
CLASS="COMMAND"
>2 BETWEEN SYMMETRIC 3 AND 1</TT
> returns
true. <TT
CLASS="COMMAND"
>BETWEEN ASYMMETRIC</TT
> was already supported.
</P
></LI
><LI
><P
> Add <TT
CLASS="COMMAND"
>NOWAIT</TT
> option to <TT
CLASS="COMMAND"
>SELECT ... FOR
UPDATE/SHARE</TT
> (Hans-Juergen Schoenig)
</P
><P
> While the <TT
CLASS="VARNAME"
>statement_timeout</TT
> configuration
parameter allows a query taking more than a certain amount of
time to be canceled, the <TT
CLASS="COMMAND"
>NOWAIT</TT
> option allows a
query to be canceled as soon as a <TT
CLASS="COMMAND"
>SELECT ... FOR
UPDATE/SHARE</TT
> command cannot immediately acquire a row lock.
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137320"
>E.169.3.4. Object Manipulation Changes</A
></H3
><P
></P
><UL
><LI
><P
> Track dependencies of shared objects (Alvaro)
</P
><P
> <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> allows global tables
(users, databases, tablespaces) to reference information in
multiple databases. This addition adds dependency information
for global tables, so, for example, user ownership can be
tracked across databases, so a user who owns something in any
database can no longer be removed. Dependency tracking already
existed for database-local objects.
</P
></LI
><LI
><P
> Allow limited <TT
CLASS="COMMAND"
>ALTER OWNER</TT
> commands to be performed
by the object owner (Stephen Frost)
</P
><P
> Prior releases allowed only superusers to change object owners.
Now, ownership can be transferred if the user executing the command
owns the object and would be able to create it as the new owner
(that is, the user is a member of the new owning role and that role
has the CREATE permission that would be needed to create the object
afresh).
</P
></LI
><LI
><P
> Add <TT
CLASS="COMMAND"
>ALTER</TT
> object <TT
CLASS="COMMAND"
>SET SCHEMA</TT
> capability
for some object types (tables, functions, types) (Bernd Helmle)
</P
><P
> This allows objects to be moved to different schemas.
</P
></LI
><LI
><P
> Add <TT
CLASS="COMMAND"
>ALTER TABLE ENABLE/DISABLE TRIGGER</TT
> to
disable triggers (Satoshi Nagayasu)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137339"
>E.169.3.5. Utility Command Changes</A
></H3
><P
></P
><UL
><LI
><P
> Allow <TT
CLASS="COMMAND"
>TRUNCATE</TT
> to truncate multiple tables in a
single command (Alvaro)
</P
><P
> Because of referential integrity checks, it is not allowed to
truncate a table that is part of a referential integrity
constraint. Using this new functionality, <TT
CLASS="COMMAND"
>TRUNCATE</TT
>
can be used to truncate such tables, if both tables involved in
a referential integrity constraint are truncated in a single
<TT
CLASS="COMMAND"
>TRUNCATE</TT
> command.
</P
></LI
><LI
><P
> Properly process carriage returns and line feeds in
<TT
CLASS="COMMAND"
>COPY CSV</TT
> mode (Andrew)
</P
><P
> In release 8.0, carriage returns and line feeds in <TT
CLASS="COMMAND"
>CSV
COPY TO</TT
> were processed in an inconsistent manner. (This was
documented on the TODO list.)
</P
></LI
><LI
><P
> Add <TT
CLASS="COMMAND"
>COPY WITH CSV HEADER</TT
> to allow a header line as
the first line in <TT
CLASS="COMMAND"
>COPY</TT
> (Andrew)
</P
><P
> This allows handling of the common <TT
CLASS="COMMAND"
>CSV</TT
> usage of
placing the column names on the first line of the data file. For
<TT
CLASS="COMMAND"
>COPY TO</TT
>, the first line contains the column names,
and for <TT
CLASS="COMMAND"
>COPY FROM</TT
>, the first line is ignored.
</P
></LI
><LI
><P
> On Windows, display better sub-second precision in
<TT
CLASS="COMMAND"
>EXPLAIN ANALYZE</TT
> (Magnus)
</P
></LI
><LI
><P
> Add trigger duration display to <TT
CLASS="COMMAND"
>EXPLAIN ANALYZE</TT
>
(Tom)
</P
><P
> Prior releases included trigger execution time as part of the
total execution time, but did not show it separately. It is now
possible to see how much time is spent in each trigger.
</P
></LI
><LI
><P
> Add support for <TT
CLASS="LITERAL"
>\x</TT
> hex escapes in <TT
CLASS="COMMAND"
>COPY</TT
>
(Sergey Ten)
</P
><P
> Previous releases only supported octal escapes.
</P
></LI
><LI
><P
> Make <TT
CLASS="COMMAND"
>SHOW ALL</TT
> include variable descriptions
(Matthias Schmidt)
</P
><P
> <TT
CLASS="COMMAND"
>SHOW</TT
> varname still only displays the variable's
value and does not include the description.
</P
></LI
><LI
><P
> Make <SPAN
CLASS="APPLICATION"
>initdb</SPAN
> create a new standard
database called <TT
CLASS="LITERAL"
>postgres</TT
>, and convert utilities to
use <TT
CLASS="LITERAL"
>postgres</TT
> rather than <TT
CLASS="LITERAL"
>template1</TT
> for
standard lookups (Dave)
</P
><P
> In prior releases, <TT
CLASS="LITERAL"
>template1</TT
> was used both as a
default connection for utilities like
<SPAN
CLASS="APPLICATION"
>createuser</SPAN
>, and as a template for
new databases. This caused <TT
CLASS="COMMAND"
>CREATE DATABASE</TT
> to
sometimes fail, because a new database cannot be created if
anyone else is in the template database. With this change, the
default connection database is now <TT
CLASS="LITERAL"
>postgres</TT
>,
meaning it is much less likely someone will be using
<TT
CLASS="LITERAL"
>template1</TT
> during <TT
CLASS="COMMAND"
>CREATE DATABASE</TT
>.
</P
></LI
><LI
><P
> Create new <SPAN
CLASS="APPLICATION"
>reindexdb</SPAN
> command-line
utility by moving <TT
CLASS="FILENAME"
>/contrib/reindexdb</TT
> into the
server (Euler Taveira de Oliveira)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137395"
>E.169.3.6. Data Type and Function Changes</A
></H3
><P
></P
><UL
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>MAX()</CODE
> and <CODE
CLASS="FUNCTION"
>MIN()</CODE
> aggregates for
array types (Koju Iijima)
</P
></LI
><LI
><P
> Fix <CODE
CLASS="FUNCTION"
>to_date()</CODE
> and <CODE
CLASS="FUNCTION"
>to_timestamp()</CODE
> to
behave reasonably when <TT
CLASS="LITERAL"
>CC</TT
> and <TT
CLASS="LITERAL"
>YY</TT
> fields
are both used (Karel Zak)
</P
><P
> If the format specification contains <TT
CLASS="LITERAL"
>CC</TT
> and a year
specification is <TT
CLASS="LITERAL"
>YYY</TT
> or longer, ignore the
<TT
CLASS="LITERAL"
>CC</TT
>. If the year specification is <TT
CLASS="LITERAL"
>YY</TT
> or
shorter, interpret <TT
CLASS="LITERAL"
>CC</TT
> as the previous century.
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>md5(bytea)</CODE
> (Abhijit Menon-Sen)
</P
><P
> <CODE
CLASS="FUNCTION"
>md5(text)</CODE
> already existed.
</P
></LI
><LI
><P
> Add support for <TT
CLASS="COMMAND"
>numeric ^ numeric</TT
> based on
<CODE
CLASS="FUNCTION"
>power(numeric, numeric)</CODE
>
</P
><P
> The function already existed, but there was no operator assigned
to it.
</P
></LI
><LI
><P
> Fix <TT
CLASS="TYPE"
>NUMERIC</TT
> modulus by properly truncating the quotient
during computation (Bruce)
</P
><P
> In previous releases, modulus for large values sometimes
returned negative results due to rounding of the quotient.
</P
></LI
><LI
><P
> Add a function <CODE
CLASS="FUNCTION"
>lastval()</CODE
> (Dennis Björklund)
</P
><P
> <CODE
CLASS="FUNCTION"
>lastval()</CODE
> is a simplified version of
<CODE
CLASS="FUNCTION"
>currval()</CODE
>. It automatically determines the proper
sequence name based on the most recent <CODE
CLASS="FUNCTION"
>nextval()</CODE
> or
<CODE
CLASS="FUNCTION"
>setval()</CODE
> call performed by the current session.
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>to_timestamp(DOUBLE PRECISION) (Michael Glaesemann)</CODE
>
</P
><P
> Converts Unix seconds since 1970 to a <TT
CLASS="TYPE"
>TIMESTAMP WITH
TIMEZONE</TT
>.
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>pg_postmaster_start_time()</CODE
> function (Euler
Taveira de Oliveira, Matthias Schmidt)
</P
></LI
><LI
><P
> Allow the full use of time zone names in <TT
CLASS="COMMAND"
>AT TIME
ZONE</TT
>, not just the short list previously available (Magnus)
</P
><P
> Previously, only a predefined list of time zone names were
supported by <TT
CLASS="COMMAND"
>AT TIME ZONE</TT
>. Now any supported time
zone name can be used, e.g.:
</P><PRE
CLASS="PROGRAMLISTING"
>SELECT CURRENT_TIMESTAMP AT TIME ZONE 'Europe/London';</PRE
><P>
In the above query, the time zone used is adjusted based on the
daylight saving time rules that were in effect on the supplied
date.
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>GREATEST()</CODE
> and <CODE
CLASS="FUNCTION"
>LEAST()</CODE
> variadic
functions (Pavel Stehule)
</P
><P
> These functions take a variable number of arguments and return
the greatest or least value among the arguments.
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>pg_column_size()</CODE
> (Mark Kirkwood)
</P
><P
> This returns storage size of a column, which might be compressed.
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>regexp_replace()</CODE
> (Atsushi Ogawa)
</P
><P
> This allows regular expression replacement, like sed. An optional
flag argument allows selection of global (replace all) and
case-insensitive modes.
</P
></LI
><LI
><P
> Fix interval division and multiplication (Bruce)
</P
><P
> Previous versions sometimes returned unjustified results, like
<TT
CLASS="COMMAND"
>'4 months'::interval / 5</TT
> returning <TT
CLASS="COMMAND"
>'1 mon
-6 days'</TT
>.
</P
></LI
><LI
><P
> Fix roundoff behavior in timestamp, time, and interval output (Tom)
</P
><P
> This fixes some cases in which the seconds field would be shown as
<TT
CLASS="LITERAL"
>60</TT
> instead of incrementing the higher-order fields.
</P
></LI
><LI
><P
> Add a separate day field to type <TT
CLASS="TYPE"
>interval</TT
> so a one day
interval can be distinguished from a 24 hour interval (Michael
Glaesemann)
</P
><P
> Days that contain a daylight saving time adjustment are not 24
hours long, but typically 23 or 25 hours. This change creates a
conceptual distinction between intervals of <SPAN
CLASS="QUOTE"
>"so many days"</SPAN
>
and intervals of <SPAN
CLASS="QUOTE"
>"so many hours"</SPAN
>. Adding
<TT
CLASS="LITERAL"
>1 day</TT
> to a timestamp now gives the same local time on
the next day even if a daylight saving time adjustment occurs
between, whereas adding <TT
CLASS="LITERAL"
>24 hours</TT
> will give a different
local time when this happens. For example, under US DST rules:
</P><PRE
CLASS="PROGRAMLISTING"
>'2005-04-03 00:00:00-05' + '1 day' = '2005-04-04 00:00:00-04'
'2005-04-03 00:00:00-05' + '24 hours' = '2005-04-04 01:00:00-04'</PRE
><P>
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>justify_days()</CODE
> and <CODE
CLASS="FUNCTION"
>justify_hours()</CODE
>
(Michael Glaesemann)
</P
><P
> These functions, respectively, adjust days to an appropriate
number of full months and days, and adjust hours to an
appropriate number of full days and hours.
</P
></LI
><LI
><P
> Move <TT
CLASS="FILENAME"
>/contrib/dbsize</TT
> into the backend, and rename
some of the functions (Dave Page, Andreas Pflug)
</P
><P
> <P
></P
></P><UL
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_tablespace_size()</CODE
>
</P
></LI
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_database_size()</CODE
>
</P
></LI
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_relation_size()</CODE
>
</P
></LI
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_total_relation_size()</CODE
>
</P
></LI
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_size_pretty()</CODE
>
</P
></LI
></UL
><P>
</P
><P
> <CODE
CLASS="FUNCTION"
>pg_total_relation_size()</CODE
> includes indexes and TOAST
tables.
</P
></LI
><LI
><P
> Add functions for read-only file access to the cluster directory
(Dave Page, Andreas Pflug)
</P
><P
> <P
></P
></P><UL
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_stat_file()</CODE
>
</P
></LI
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_read_file()</CODE
>
</P
></LI
><LI
><P
> <CODE
CLASS="FUNCTION"
>pg_ls_dir()</CODE
>
</P
></LI
></UL
><P>
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>pg_reload_conf()</CODE
> to force reloading of the
configuration files (Dave Page, Andreas Pflug)
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>pg_rotate_logfile()</CODE
> to force rotation of the
server log file (Dave Page, Andreas Pflug)
</P
></LI
><LI
><P
> Change <TT
CLASS="LITERAL"
>pg_stat_*</TT
> views to include TOAST tables (Tom)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137530"
>E.169.3.7. Encoding and Locale Changes</A
></H3
><P
></P
><UL
><LI
><P
> Rename some encodings to be more consistent and to follow
international standards (Bruce)
</P
><P
> <P
></P
></P><UL
><LI
><P
> <TT
CLASS="LITERAL"
>UNICODE</TT
> is now <TT
CLASS="LITERAL"
>UTF8</TT
>
</P
></LI
><LI
><P
> <TT
CLASS="LITERAL"
>ALT</TT
> is now <TT
CLASS="LITERAL"
>WIN866</TT
>
</P
></LI
><LI
><P
> <TT
CLASS="LITERAL"
>WIN</TT
> is now <TT
CLASS="LITERAL"
>WIN1251</TT
>
</P
></LI
><LI
><P
> <TT
CLASS="LITERAL"
>TCVN</TT
> is now <TT
CLASS="LITERAL"
>WIN1258</TT
>
</P
></LI
></UL
><P>
</P
><P
> The original names still work.
</P
></LI
><LI
><P
> Add support for <TT
CLASS="LITERAL"
>WIN1252</TT
> encoding (Roland Volkmann)
</P
></LI
><LI
><P
> Add support for four-byte <TT
CLASS="LITERAL"
>UTF8</TT
> characters (John
Hansen)
</P
><P
> Previously only one, two, and three-byte <TT
CLASS="LITERAL"
>UTF8</TT
> characters
were supported. This is particularly important for support for
some Chinese character sets.
</P
></LI
><LI
><P
> Allow direct conversion between <TT
CLASS="LITERAL"
>EUC_JP</TT
> and
<TT
CLASS="LITERAL"
>SJIS</TT
> to improve performance (Atsushi Ogawa)
</P
></LI
><LI
><P
> Allow the UTF8 encoding to work on Windows (Magnus)
</P
><P
> This is done by mapping UTF8 to the Windows-native UTF16
implementation.
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137569"
>E.169.3.8. General Server-Side Language Changes</A
></H3
><P
></P
><UL
><LI
><P
> Fix <TT
CLASS="COMMAND"
>ALTER LANGUAGE RENAME</TT
> (Sergey Yatskevich)
</P
></LI
><LI
><P
> Allow function characteristics, like strictness and volatility,
to be modified via <TT
CLASS="COMMAND"
>ALTER FUNCTION</TT
> (Neil)
</P
></LI
><LI
><P
> Increase the maximum number of function arguments to 100 (Tom)
</P
></LI
><LI
><P
> Allow SQL and PL/pgSQL functions to use <TT
CLASS="COMMAND"
>OUT</TT
> and
<TT
CLASS="COMMAND"
>INOUT</TT
> parameters (Tom)
</P
><P
> <TT
CLASS="COMMAND"
>OUT</TT
> is an alternate way for a function to return
values. Instead of using <TT
CLASS="COMMAND"
>RETURN</TT
>, values can be
returned by assigning to parameters declared as <TT
CLASS="COMMAND"
>OUT</TT
> or
<TT
CLASS="COMMAND"
>INOUT</TT
>. This is notationally simpler in some cases,
particularly so when multiple values need to be returned.
While returning multiple values from a function
was possible in previous releases, this greatly simplifies the
process. (The feature will be extended to other server-side
languages in future releases.)
</P
></LI
><LI
><P
> Move language handler functions into the <TT
CLASS="LITERAL"
>pg_catalog</TT
> schema
</P
><P
> This makes it easier to drop the public schema if desired.
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>SPI_getnspname()</CODE
> to SPI (Neil)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137596"
>E.169.3.9. PL/pgSQL Server-Side Language Changes</A
></H3
><P
></P
><UL
><LI
><P
> Overhaul the memory management of PL/pgSQL functions (Neil)
</P
><P
> The parsetree of each function is now stored in a separate
memory context. This allows this memory to be easily reclaimed
when it is no longer needed.
</P
></LI
><LI
><P
> Check function syntax at <TT
CLASS="COMMAND"
>CREATE FUNCTION</TT
> time,
rather than at runtime (Neil)
</P
><P
> Previously, most syntax errors were reported only when the
function was executed.
</P
></LI
><LI
><P
> Allow <TT
CLASS="COMMAND"
>OPEN</TT
> to open non-<TT
CLASS="COMMAND"
>SELECT</TT
> queries
like <TT
CLASS="COMMAND"
>EXPLAIN</TT
> and <TT
CLASS="COMMAND"
>SHOW</TT
> (Tom)
</P
></LI
><LI
><P
> No longer require functions to issue a <TT
CLASS="COMMAND"
>RETURN</TT
>
statement (Tom)
</P
><P
> This is a byproduct of the newly added <TT
CLASS="COMMAND"
>OUT</TT
> and
<TT
CLASS="COMMAND"
>INOUT</TT
> functionality. <TT
CLASS="COMMAND"
>RETURN</TT
> can
be omitted when it is not needed to provide the function's
return value.
</P
></LI
><LI
><P
> Add support for an optional <TT
CLASS="COMMAND"
>INTO</TT
> clause to
PL/pgSQL's <TT
CLASS="COMMAND"
>EXECUTE</TT
> statement (Pavel Stehule, Neil)
</P
></LI
><LI
><P
> Make <TT
CLASS="COMMAND"
>CREATE TABLE AS</TT
> set <TT
CLASS="COMMAND"
>ROW_COUNT</TT
> (Tom)
</P
></LI
><LI
><P
> Define <TT
CLASS="LITERAL"
>SQLSTATE</TT
> and <TT
CLASS="LITERAL"
>SQLERRM</TT
> to return
the <TT
CLASS="LITERAL"
>SQLSTATE</TT
> and error message of the current
exception (Pavel Stehule, Neil)
</P
><P
> These variables are only defined inside exception blocks.
</P
></LI
><LI
><P
> Allow the parameters to the <TT
CLASS="COMMAND"
>RAISE</TT
> statement to be
expressions (Pavel Stehule, Neil)
</P
></LI
><LI
><P
> Add a loop <TT
CLASS="COMMAND"
>CONTINUE</TT
> statement (Pavel Stehule, Neil)
</P
></LI
><LI
><P
> Allow block and loop labels (Pavel Stehule)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137641"
>E.169.3.10. PL/Perl Server-Side Language Changes</A
></H3
><P
></P
><UL
><LI
><P
> Allow large result sets to be returned efficiently (Abhijit
Menon-Sen)
</P
><P
> This allows functions to use <CODE
CLASS="FUNCTION"
>return_next()</CODE
> to avoid
building the entire result set in memory.
</P
></LI
><LI
><P
> Allow one-row-at-a-time retrieval of query results (Abhijit Menon-Sen)
</P
><P
> This allows functions to use <CODE
CLASS="FUNCTION"
>spi_query()</CODE
> and
<CODE
CLASS="FUNCTION"
>spi_fetchrow()</CODE
> to avoid accumulating the entire
result set in memory.
</P
></LI
><LI
><P
> Force PL/Perl to handle strings as <TT
CLASS="LITERAL"
>UTF8</TT
> if the
server encoding is <TT
CLASS="LITERAL"
>UTF8</TT
> (David Kamholz)
</P
></LI
><LI
><P
> Add a validator function for PL/Perl (Andrew)
</P
><P
> This allows syntax errors to be reported at definition time,
rather than execution time.
</P
></LI
><LI
><P
> Allow PL/Perl to return a Perl array when the function returns
an array type (Andrew)
</P
><P
> This basically maps <SPAN
CLASS="PRODUCTNAME"
>PostgreSQL</SPAN
> arrays
to Perl arrays.
</P
></LI
><LI
><P
> Allow Perl nonfatal warnings to generate <TT
CLASS="COMMAND"
>NOTICE</TT
>
messages (Andrew)
</P
></LI
><LI
><P
> Allow Perl's <TT
CLASS="LITERAL"
>strict</TT
> mode to be enabled (Andrew)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137670"
>E.169.3.11. <SPAN
CLASS="APPLICATION"
>psql</SPAN
> Changes</A
></H3
><P
></P
><UL
><LI
><P
> Add <TT
CLASS="COMMAND"
>\set ON_ERROR_ROLLBACK</TT
> to allow statements in
a transaction to error without affecting the rest of the
transaction (Greg Sabino Mullane)
</P
><P
> This is basically implemented by wrapping every statement in a
sub-transaction.
</P
></LI
><LI
><P
> Add support for <TT
CLASS="LITERAL"
>\x</TT
> hex strings in
<SPAN
CLASS="APPLICATION"
>psql</SPAN
> variables (Bruce)
</P
><P
> Octal escapes were already supported.
</P
></LI
><LI
><P
> Add support for <TT
CLASS="COMMAND"
>troff -ms</TT
> output format (Roger
Leigh)
</P
></LI
><LI
><P
> Allow the history file location to be controlled by
<TT
CLASS="ENVAR"
>HISTFILE</TT
> (Andreas Seltenreich)
</P
><P
> This allows configuration of per-database history storage.
</P
></LI
><LI
><P
> Prevent <TT
CLASS="COMMAND"
>\x</TT
> (expanded mode) from affecting
the output of <TT
CLASS="COMMAND"
>\d tablename</TT
> (Neil)
</P
></LI
><LI
><P
> Add <TT
CLASS="OPTION"
>-L</TT
> option to <SPAN
CLASS="APPLICATION"
>psql</SPAN
> to
log sessions (Lorne Sunley)
</P
><P
> This option was added because some operating systems do not have
simple command-line activity logging functionality.
</P
></LI
><LI
><P
> Make <TT
CLASS="COMMAND"
>\d</TT
> show the tablespaces of indexes (Qingqing
Zhou)
</P
></LI
><LI
><P
> Allow <SPAN
CLASS="APPLICATION"
>psql</SPAN
> help (<TT
CLASS="COMMAND"
>\h</TT
>) to
make a best guess on the proper help information (Greg Sabino
Mullane)
</P
><P
> This allows the user to just add <TT
CLASS="COMMAND"
>\h</TT
> to the front of
the syntax error query and get help on the supported syntax.
Previously any additional query text beyond the command name
had to be removed to use <TT
CLASS="COMMAND"
>\h</TT
>.
</P
></LI
><LI
><P
> Add <TT
CLASS="COMMAND"
>\pset numericlocale</TT
> to allow numbers to be
output in a locale-aware format (Eugen Nedelcu)
</P
><P
> For example, using <TT
CLASS="LITERAL"
>C</TT
> locale <TT
CLASS="LITERAL"
>100000</TT
> would
be output as <TT
CLASS="LITERAL"
>100,000.0</TT
> while a European locale might
output this value as <TT
CLASS="LITERAL"
>100.000,0</TT
>.
</P
></LI
><LI
><P
> Make startup banner show both server version number and
<SPAN
CLASS="APPLICATION"
>psql</SPAN
>'s version number, when they are different (Bruce)
</P
><P
> Also, a warning will be shown if the server and <SPAN
CLASS="APPLICATION"
>psql</SPAN
>
are from different major releases.
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137722"
>E.169.3.12. <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> Changes</A
></H3
><P
></P
><UL
><LI
><P
> Add <TT
CLASS="OPTION"
>-n</TT
> / <TT
CLASS="OPTION"
>--schema</TT
> switch to
<SPAN
CLASS="APPLICATION"
>pg_restore</SPAN
> (Richard van den Berg)
</P
><P
> This allows just the objects in a specified schema to be restored.
</P
></LI
><LI
><P
> Allow <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> to dump large objects even in
text mode (Tom)
</P
><P
> With this change, large objects are now always dumped; the former
<TT
CLASS="OPTION"
>-b</TT
> switch is a no-op.
</P
></LI
><LI
><P
> Allow <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> to dump a consistent snapshot of
large objects (Tom)
</P
></LI
><LI
><P
> Dump comments for large objects (Tom)
</P
></LI
><LI
><P
> Add <TT
CLASS="OPTION"
>--encoding</TT
> to <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
>
(Magnus Hagander)
</P
><P
> This allows a database to be dumped in an encoding that is
different from the server's encoding. This is valuable when
transferring the dump to a machine with a different encoding.
</P
></LI
><LI
><P
> Rely on <TT
CLASS="STRUCTNAME"
>pg_pltemplate</TT
> for procedural languages (Tom)
</P
><P
> If the call handler for a procedural language is in the
<TT
CLASS="LITERAL"
>pg_catalog</TT
> schema, <SPAN
CLASS="APPLICATION"
>pg_dump</SPAN
> does not
dump the handler. Instead, it dumps the language using just
<TT
CLASS="COMMAND"
>CREATE LANGUAGE <TT
CLASS="REPLACEABLE"
><I
>name</I
></TT
></TT
>,
relying on the <TT
CLASS="STRUCTNAME"
>pg_pltemplate</TT
> catalog to provide
the language's creation parameters at load time.
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137756"
>E.169.3.13. <SPAN
CLASS="APPLICATION"
>libpq</SPAN
> Changes</A
></H3
><P
></P
><UL
><LI
><P
> Add a <TT
CLASS="ENVAR"
>PGPASSFILE</TT
> environment variable to specify the
password file's filename (Andrew)
</P
></LI
><LI
><P
> Add <CODE
CLASS="FUNCTION"
>lo_create()</CODE
>, that is similar to
<CODE
CLASS="FUNCTION"
>lo_creat()</CODE
> but allows the OID of the large object
to be specified (Tom)
</P
></LI
><LI
><P
> Make <SPAN
CLASS="APPLICATION"
>libpq</SPAN
> consistently return an error
to the client application on <CODE
CLASS="FUNCTION"
>malloc()</CODE
>
failure (Neil)
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137771"
>E.169.3.14. Source Code Changes</A
></H3
><P
></P
><UL
><LI
><P
> Fix <SPAN
CLASS="APPLICATION"
>pgxs</SPAN
> to support building against a relocated
installation
</P
></LI
><LI
><P
> Add spinlock support for the Itanium processor using Intel
compiler (Vikram Kalsi)
</P
></LI
><LI
><P
> Add Kerberos 5 support for Windows (Magnus)
</P
></LI
><LI
><P
> Add Chinese FAQ (laser@pgsqldb.com)
</P
></LI
><LI
><P
> Rename Rendezvous to Bonjour to match OS/X feature renaming
(Bruce)
</P
></LI
><LI
><P
> Add support for <TT
CLASS="LITERAL"
>fsync_writethrough</TT
> on
Darwin (Chris Campbell)
</P
></LI
><LI
><P
> Streamline the passing of information within the server, the
optimizer, and the lock system (Tom)
</P
></LI
><LI
><P
> Allow <SPAN
CLASS="APPLICATION"
>pg_config</SPAN
> to be compiled using MSVC (Andrew)
</P
><P
> This is required to build DBD::Pg using <SPAN
CLASS="APPLICATION"
>MSVC</SPAN
>.
</P
></LI
><LI
><P
> Remove support for Kerberos V4 (Magnus)
</P
><P
> Kerberos 4 had security vulnerabilities and is no longer
maintained.
</P
></LI
><LI
><P
> Code cleanups (Coverity static analysis performed by
EnterpriseDB)
</P
></LI
><LI
><P
> Modify <TT
CLASS="FILENAME"
>postgresql.conf</TT
> to use documentation defaults
<TT
CLASS="LITERAL"
>on</TT
>/<TT
CLASS="LITERAL"
>off</TT
> rather than
<TT
CLASS="LITERAL"
>true</TT
>/<TT
CLASS="LITERAL"
>false</TT
> (Bruce)
</P
></LI
><LI
><P
> Enhance <SPAN
CLASS="APPLICATION"
>pg_config</SPAN
> to be able to report more
build-time values (Tom)
</P
></LI
><LI
><P
> Allow <SPAN
CLASS="APPLICATION"
>libpq</SPAN
> to be built thread-safe
on Windows (Dave Page)
</P
></LI
><LI
><P
> Allow IPv6 connections to be used on Windows (Andrew)
</P
></LI
><LI
><P
> Add Server Administration documentation about I/O subsystem
reliability (Bruce)
</P
></LI
><LI
><P
> Move private declarations from <TT
CLASS="FILENAME"
>gist.h</TT
> to
<TT
CLASS="FILENAME"
>gist_private.h</TT
> (Neil)
</P
><P
> In previous releases, <TT
CLASS="FILENAME"
>gist.h</TT
> contained both the
public GiST API (intended for use by authors of GiST index
implementations) as well as some private declarations used by
the implementation of GiST itself. The latter have been moved
to a separate file, <TT
CLASS="FILENAME"
>gist_private.h</TT
>. Most GiST
index implementations should be unaffected.
</P
></LI
><LI
><P
> Overhaul GiST memory management (Neil)
</P
><P
> GiST methods are now always invoked in a short-lived memory
context. Therefore, memory allocated via <CODE
CLASS="FUNCTION"
>palloc()</CODE
>
will be reclaimed automatically, so GiST index implementations
do not need to manually release allocated memory via
<CODE
CLASS="FUNCTION"
>pfree()</CODE
>.
</P
></LI
></UL
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN137829"
>E.169.3.15. Contrib Changes</A
></H3
><P
></P
><UL
><LI
><P
> Add <TT
CLASS="FILENAME"
>/contrib/pg_buffercache</TT
> contrib module (Mark
Kirkwood)
</P
><P
> This displays the contents of the buffer cache, for debugging and
performance tuning purposes.
</P
></LI
><LI
><P
> Remove <TT
CLASS="FILENAME"
>/contrib/array</TT
> because it is obsolete (Tom)
</P
></LI
><LI
><P
> Clean up the <TT
CLASS="FILENAME"
>/contrib/lo</TT
> module (Tom)
</P
></LI
><LI
><P
> Move <TT
CLASS="FILENAME"
>/contrib/findoidjoins</TT
> to
<TT
CLASS="FILENAME"
>/src/tools</TT
> (Tom)
</P
></LI
><LI
><P
> Remove the <TT
CLASS="LITERAL"
><<</TT
>, <TT
CLASS="LITERAL"
>>></TT
>,
<TT
CLASS="LITERAL"
>&<</TT
>, and <TT
CLASS="LITERAL"
>&></TT
> operators from
<TT
CLASS="FILENAME"
>/contrib/cube</TT
>
</P
><P
> These operators were not useful.
</P
></LI
><LI
><P
> Improve <TT
CLASS="FILENAME"
>/contrib/btree_gist</TT
> (Janko Richter)
</P
></LI
><LI
><P
> Improve <TT
CLASS="FILENAME"
>/contrib/pgbench</TT
> (Tomoaki Sato, Tatsuo)
</P
><P
> There is now a facility for testing with SQL command scripts given
by the user, instead of only a hard-wired command sequence.
</P
></LI
><LI
><P
> Improve <TT
CLASS="FILENAME"
>/contrib/pgcrypto</TT
> (Marko Kreen)
</P
><P
></P
><UL
><LI
><P
> Implementation of OpenPGP symmetric-key and public-key encryption
</P
><P
> Both RSA and Elgamal public-key algorithms are supported.
</P
></LI
><LI
><P
> Stand alone build: include SHA256/384/512 hashes, Fortuna PRNG
</P
></LI
><LI
><P
> OpenSSL build: support 3DES, use internal AES with OpenSSL < 0.9.7
</P
></LI
><LI
><P
> Take build parameters (OpenSSL, zlib) from <TT
CLASS="FILENAME"
>configure</TT
> result
</P
><P
> There is no need to edit the <TT
CLASS="FILENAME"
>Makefile</TT
> anymore.
</P
></LI
><LI
><P
> Remove support for <TT
CLASS="FILENAME"
>libmhash</TT
> and <TT
CLASS="FILENAME"
>libmcrypt</TT
>
</P
></LI
></UL
></LI
></UL
></DIV
></DIV
></DIV
><DIV
CLASS="NAVFOOTER"
><HR
ALIGN="LEFT"
WIDTH="100%"><TABLE
SUMMARY="Footer navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
><A
HREF="release-8-1-1.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="index.html"
ACCESSKEY="H"
>Home</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
><A
HREF="release-8-0-26.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
>Release 8.1.1</TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="release.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
>Release 8.0.26</TD
></TR
></TABLE
></DIV
></BODY
></HTML
>