rdelaage
/
postgresql
mirror of https://git.postgresql.org/git/postgresql.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Editorial improvements for GIN documentation.

This commit is contained in:
Tom Lane 2006-12-01 23:46:46 +00:00
parent a88ec7b4bb
commit 08fa6a6851
3 changed files with 138 additions and 114 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

180
doc/src/sgml/gin.sgml
View File

@ -1,4 +1,4 @@
<!-- $PostgreSQL: pgsql/doc/src/sgml/gin.sgml,v 2.6 2006/11/30 20:50:44 petere Exp $ --> <!-- $PostgreSQL: pgsql/doc/src/sgml/gin.sgml,v 2.7 2006/12/01 23:46:46 tgl Exp $ -->
<chapter id="GIN"> <chapter id="GIN">
<title>GIN Indexes</title> <title>GIN Indexes</title>
@ -14,8 +14,9 @@
<para> <para>
<acronym>GIN</acronym> stands for Generalized Inverted Index. It is <acronym>GIN</acronym> stands for Generalized Inverted Index. It is
an index structure storing a set of (key, posting list) pairs, where an index structure storing a set of (key, posting list) pairs, where
'posting list' is a set of rows in which the key occurs. Each a <quote>posting list</> is a set of rows in which the key occurs. Each
row may contain many keys. indexed value may contain many keys, so the same row ID may appear in
multiple posting lists.
</para> </para>
<para> <para>
@ -45,7 +46,7 @@
<para> <para>
The <acronym>GIN</acronym> interface has a high level of abstraction, The <acronym>GIN</acronym> interface has a high level of abstraction,
requiring the access method implementer to only implement the semantics of requiring the access method implementer only to implement the semantics of
the data type being accessed. The <acronym>GIN</acronym> layer itself the data type being accessed. The <acronym>GIN</acronym> layer itself
takes care of concurrency, logging and searching the tree structure. takes care of concurrency, logging and searching the tree structure.
</para> </para>
@ -53,26 +54,14 @@
<para> <para>
All it takes to get a <acronym>GIN</acronym> access method working All it takes to get a <acronym>GIN</acronym> access method working
is to implement four user-defined methods, which define the behavior of is to implement four user-defined methods, which define the behavior of
keys in the tree. In short, <acronym>GIN</acronym> combines extensibility keys in the tree and the relationships between keys, indexed values,
along with generality, code reuse, and a clean interface. and indexable queries. In short, <acronym>GIN</acronym> combines
</para> extensibility with generality, code reuse, and a clean interface.
</sect1>
<sect1 id="gin-implementation">
<title>Implementation</title>
<para>
Internally, <acronym>GIN</acronym> consists of a B-tree index constructed
over keys, where each key is an element of the indexed value
(element of array, for example) and where each tuple in a leaf page is
either a pointer to a B-tree over heap pointers (PT, posting tree), or a
list of heap pointers (PL, posting list) if the tuple is small enough.
</para> </para>
<para> <para>
There are four methods that an index operator class for The four methods that an index operator class for
<acronym>GIN</acronym> must provide (prototypes are in pseudocode): <acronym>GIN</acronym> must provide are:
</para> </para>
<variablelist> <variablelist>
@ -80,9 +69,9 @@
<term>int compare(Datum a, Datum b)</term> <term>int compare(Datum a, Datum b)</term>
<listitem> <listitem>
<para> <para>
Compares keys (not indexed values!) and returns an integer less than Compares keys (not indexed values!) and returns an integer less than
zero, zero, or greater than zero, indicating whether the first key is zero, zero, or greater than zero, indicating whether the first key is
less than, equal to, or greater than the second. less than, equal to, or greater than the second.
</para> </para>
</listitem> </listitem>
</varlistentry> </varlistentry>
@ -91,21 +80,26 @@
<term>Datum* extractValue(Datum inputValue, uint32 *nkeys)</term> <term>Datum* extractValue(Datum inputValue, uint32 *nkeys)</term>
<listitem> <listitem>
<para> <para>
Returns an array of keys of value to be indexed, nkeys should Returns an array of keys given a value to be indexed. The
contain the number of returned keys. number of returned keys must be stored into <literal>*nkeys</>.
</para> </para>
</listitem> </listitem>
</varlistentry> </varlistentry>
<varlistentry> <varlistentry>
<term>Datum* extractQuery(Datum query, uint32 nkeys, <term>Datum* extractQuery(Datum query, uint32 *nkeys,
StrategyNumber n)</term> StrategyNumber n)</term>
<listitem> <listitem>
<para> <para>
Returns an array of keys of the query to be executed. n contains the Returns an array of keys given a value to be queried; that is,
strategy number of the operation (see <xref <literal>query</> is the value on the right-hand side of an
linkend="xindex-strategies">). Depending on n, query may be indexable operator whose left-hand side is the indexed column.
different type. <literal>n</> is the strategy number of the operator within the
operator class (see <xref linkend="xindex-strategies">).
Often, <function>extractQuery</> will need
to consult <literal>n</> to determine the data type of
<literal>query</> and the key values that need to be extracted.
The number of returned keys must be stored into <literal>*nkeys</>.
</para> </para>
</listitem> </listitem>
</varlistentry> </varlistentry>
@ -114,11 +108,16 @@
<term>bool consistent(bool check[], StrategyNumber n, Datum query)</term> <term>bool consistent(bool check[], StrategyNumber n, Datum query)</term>
<listitem> <listitem>
<para> <para>
Returns TRUE if the indexed value satisfies the query qualifier with Returns TRUE if the indexed value satisfies the query operator with
strategy n (or may satisfy in case of RECHECK mark in operator class). strategy number <literal>n</> (or may satisfy, if the operator is
Each element of the check array is TRUE if the indexed value has a marked RECHECK in the operator class). The <literal>check</> array has
corresponding key in the query: if (check[i] == TRUE) the i-th key of the same length as the number of keys previously returned by
the query is present in the indexed value. <function>extractQuery</> for this query. Each element of the
<literal>check</> array is TRUE if the indexed value contains the
corresponding query key, ie, if (check[i] == TRUE) the i-th key of the
<function>extractQuery</> result array is present in the indexed value.
The original <literal>query</> datum (not the extracted key array!) is
passed in case the <function>consistent</> method needs to consult it.
</para> </para>
</listitem> </listitem>
</varlistentry> </varlistentry>
@ -127,6 +126,19 @@
</sect1> </sect1>
<sect1 id="gin-implementation">
<title>Implementation</title>
<para>
Internally, a <acronym>GIN</acronym> index contains a B-tree index
constructed over keys, where each key is an element of the indexed value
(a member of an array, for example) and where each tuple in a leaf page is
either a pointer to a B-tree over heap pointers (PT, posting tree), or a
list of heap pointers (PL, posting list) if the list is small enough.
</para>
</sect1>
<sect1 id="gin-tips"> <sect1 id="gin-tips">
<title>GIN tips and tricks</title> <title>GIN tips and tricks</title>
@ -134,44 +146,43 @@
<varlistentry> <varlistentry>
<term>Create vs insert</term> <term>Create vs insert</term>
<listitem> <listitem>
<para> <para>
In most cases, insertion into a <acronym>GIN</acronym> index is slow In most cases, insertion into a <acronym>GIN</acronym> index is slow
due to the likelihood of many keys being inserted for each value. due to the likelihood of many keys being inserted for each value.
So, for bulk insertions into a table it is advisable to to drop the GIN So, for bulk insertions into a table it is advisable to drop the GIN
index and recreate it after finishing bulk insertion. index and recreate it after finishing bulk insertion.
</para> </para>
</listitem> </listitem>
</varlistentry> </varlistentry>
<varlistentry> <varlistentry>
<term>gin_fuzzy_search_limit</term> <term><xref linkend="guc-gin-fuzzy-search-limit"></term>
<listitem> <listitem>
<para> <para>
The primary goal of developing <acronym>GIN</acronym> indices was The primary goal of developing <acronym>GIN</acronym> indexes was
support for highly scalable, full-text search in to create support for highly scalable, full-text search in
<productname>PostgreSQL</productname> and there are often situations when <productname>PostgreSQL</productname>, and there are often situations when
a full-text search returns a very large set of results. Since reading a full-text search returns a very large set of results. Moreover, this
tuples from the disk and sorting them could take a lot of time, this is often happens when the query contains very frequent words, so that the
unacceptable for production. (Note that the index search itself is very large result set is not even useful. Since reading many
fast.) tuples from the disk and sorting them could take a lot of time, this is
unacceptable for production. (Note that the index search itself is very
fast.)
</para>
<para>
To facilitate controlled execution of such queries
<acronym>GIN</acronym> has a configurable soft upper limit on the size
of the returned set, the
<varname>gin_fuzzy_search_limit</varname> configuration parameter.
It is set to 0 (meaning no limit) by default.
If a non-zero limit is set, then the returned set is a subset of
the whole result set, chosen at random.
</para>
<para>
<quote>Soft</quote> means that the actual number of returned results
could differ slightly from the specified limit, depending on the query
and the quality of the system's random number generator.
</para> </para>
<para>
Such queries usually contain very frequent words, so the results are not
very helpful. To facilitate execution of such queries
<acronym>GIN</acronym> has a configurable soft upper limit of the size
of the returned set, determined by the
<varname>gin_fuzzy_search_limit</varname> GUC variable. It is set to 0 by
default (no limit).
</para>
<para>
If a non-zero search limit is set, then the returned set is a subset of
the whole result set, chosen at random.
</para>
<para>
<quote>Soft</quote> means that the actual number of returned results
could slightly differ from the specified limit, depending on the query
and the quality of the system's random number generator.
</para>
</listitem> </listitem>
</varlistentry> </varlistentry>
</variablelist> </variablelist>
@ -182,21 +193,30 @@
<title>Limitations</title> <title>Limitations</title>
<para> <para>
<acronym>GIN</acronym> doesn't support full index scans due to their <acronym>GIN</acronym> doesn't support full index scans: because there are
extreme inefficiency: because there are often many keys per value, often many keys per value, each heap pointer would be returned many times,
each heap pointer will be returned several times. and there is no easy way to prevent this.
</para> </para>
<para> <para>
When <function>extractQuery</function> returns zero keys, When <function>extractQuery</function> returns zero keys,
<acronym>GIN</acronym> will emit an error: for different opclasses and <acronym>GIN</acronym> will emit an error. Depending on the operator,
strategies the semantic meaning of a void query may be different (for a void query might match all, some, or none of the indexed values (for
example, any array contains the void array, but they don't overlap the example, every array contains the empty array, but does not overlap the
void array), and <acronym>GIN</acronym> can't suggest a reasonable answer. empty array), and <acronym>GIN</acronym> can't determine the correct
answer, nor produce a full-index-scan result if it could determine that
that was correct.
</para> </para>
<para> <para>
<acronym>GIN</acronym> searches keys only by equality matching. This may It is not an error for <function>extractValue</> to return zero keys,
but in this case the indexed value will be unrepresented in the index.
This is another reason why full index scan is not useful &mdash; it would
miss such rows.
</para>
<para>
<acronym>GIN</acronym> searches keys only by equality matching. This may
be improved in future. be improved in future.
</para> </para>
</sect1> </sect1>
@ -206,12 +226,12 @@
<para> <para>
The <productname>PostgreSQL</productname> source distribution includes The <productname>PostgreSQL</productname> source distribution includes
<acronym>GIN</acronym> classes for one-dimensional arrays of all internal <acronym>GIN</acronym> classes for one-dimensional arrays of all internal
types. The following types. The following
<filename>contrib</> modules also contain <acronym>GIN</acronym> <filename>contrib</> modules also contain <acronym>GIN</acronym>
operator classes: operator classes:
</para> </para>
<variablelist> <variablelist>
<varlistentry> <varlistentry>
<term>intarray</term> <term>intarray</term>

11
doc/src/sgml/indices.sgml
View File

@ -1,4 +1,4 @@
<!-- $PostgreSQL: pgsql/doc/src/sgml/indices.sgml,v 1.65 2006/10/23 18:10:31 petere Exp $ --> <!-- $PostgreSQL: pgsql/doc/src/sgml/indices.sgml,v 1.66 2006/12/01 23:46:46 tgl Exp $ -->
<chapter id="indexes"> <chapter id="indexes">
<title id="indexes-title">Indexes</title> <title id="indexes-title">Indexes</title>
@ -116,7 +116,7 @@ CREATE INDEX test1_id_index ON test1 (id);
<para> <para>
<productname>PostgreSQL</productname> provides several index types: <productname>PostgreSQL</productname> provides several index types:
B-tree, Hash, GIN and GiST. Each index type uses a different B-tree, Hash, GiST and GIN. Each index type uses a different
algorithm that is best suited to different types of queries. algorithm that is best suited to different types of queries.
By default, the <command>CREATE INDEX</command> command will create a By default, the <command>CREATE INDEX</command> command will create a
B-tree index, which fits the most common situations. B-tree index, which fits the most common situations.
@ -247,8 +247,8 @@ CREATE INDEX <replaceable>name</replaceable> ON <replaceable>table</replaceable>
<primary>GIN</primary> <primary>GIN</primary>
<see>index</see> <see>index</see>
</indexterm> </indexterm>
GIN is a inverted index and it's usable for values which have more GIN indexes are inverted indexes which can handle values that contain more
than one key, arrays for example. Like GiST, GIN may support than one key, arrays for example. Like GiST, GIN can support
many different user-defined indexing strategies and the particular many different user-defined indexing strategies and the particular
operators with which a GIN index can be used vary depending on the operators with which a GIN index can be used vary depending on the
indexing strategy. indexing strategy.
@ -267,7 +267,8 @@ CREATE INDEX <replaceable>name</replaceable> ON <replaceable>table</replaceable>
(See <xref linkend="functions-array"> for the meaning of (See <xref linkend="functions-array"> for the meaning of
these operators.) these operators.)
Other GIN operator classes are available in the <literal>contrib</> Other GIN operator classes are available in the <literal>contrib</>
<literal>tsearch2</literal> and <literal>intarray</literal> modules. For more information see <xref linkend="GIN">. <literal>tsearch2</literal> and <literal>intarray</literal> modules.
For more information see <xref linkend="GIN">.
</para> </para>
</sect1> </sect1>

61
doc/src/sgml/xindex.sgml
View File

@ -1,4 +1,4 @@
<!-- $PostgreSQL: pgsql/doc/src/sgml/xindex.sgml,v 1.52 2006/10/23 18:10:32 petere Exp $ --> <!-- $PostgreSQL: pgsql/doc/src/sgml/xindex.sgml,v 1.53 2006/12/01 23:46:46 tgl Exp $ -->
<sect1 id="xindex"> <sect1 id="xindex">
<title>Interfacing Extensions To Indexes</title> <title>Interfacing Extensions To Indexes</title>
@ -243,15 +243,16 @@
</table> </table>
<para> <para>
GIN indexes are similar to GiST's in flexibility: they don't have a fixed GIN indexes are similar to GiST indexes in flexibility: they don't have a
et of strategies. Instead, the <quote>consistency</> support routine fixed set of strategies. Instead the support routines of each operator
interprets the strategy numbers accordingly with operator class class interpret the strategy numbers according to the operator class's
definition. As an example, strategies of operator class over arrays definition. As an example, the strategy numbers used by the built-in
is shown in <xref linkend="xindex-gin-array-strat-table">. operator classes for arrays are
shown in <xref linkend="xindex-gin-array-strat-table">.
</para> </para>
<table tocentry="1" id="xindex-gin-array-strat-table"> <table tocentry="1" id="xindex-gin-array-strat-table">
<title>GIN Array's Strategies</title> <title>GIN Array Strategies</title>
<tgroup cols="2"> <tgroup cols="2">
<thead> <thead>
<row> <row>
@ -388,36 +389,35 @@
<tbody> <tbody>
<row> <row>
<entry>consistent - determine whether key satisfies the <entry>consistent - determine whether key satisfies the
query qualifier</entry> query qualifier</entry>
<entry>1</entry> <entry>1</entry>
</row> </row>
<row> <row>
<entry>union - compute union of of a set of given keys</entry> <entry>union - compute union of a set of keys</entry>
<entry>2</entry> <entry>2</entry>
</row> </row>
<row> <row>
<entry>compress - computes a compressed representation of a key or value <entry>compress - compute a compressed representation of a key or value
to be indexed</entry> to be indexed</entry>
<entry>3</entry> <entry>3</entry>
</row> </row>
<row> <row>
<entry>decompress - computes a decompressed representation of a <entry>decompress - compute a decompressed representation of a
compressed key </entry> compressed key</entry>
<entry>4</entry> <entry>4</entry>
</row> </row>
<row> <row>
<entry>penalty - compute penalty for inserting new key into subtree <entry>penalty - compute penalty for inserting new key into subtree
with given subtree's key</entry> with given subtree's key</entry>
<entry>5</entry> <entry>5</entry>
</row> </row>
<row> <row>
<entry>picksplit - determine which entries of a page are to be moved <entry>picksplit - determine which entries of a page are to be moved
to the new page and compute the union keys for resulting pages </entry> to the new page and compute the union keys for resulting pages</entry>
<entry>6</entry> <entry>6</entry>
</row> </row>
<row> <row>
<entry>equal - compare two keys and returns true if they are equal <entry>equal - compare two keys and return true if they are equal</entry>
</entry>
<entry>7</entry> <entry>7</entry>
</row> </row>
</tbody> </tbody>
@ -441,23 +441,22 @@
<tbody> <tbody>
<row> <row>
<entry> <entry>
compare - Compare two keys and return an integer less than zero, zero, or compare - compare two keys and return an integer less than zero, zero,
greater than zero, indicating whether the first key is less than, equal to, or greater than zero, indicating whether the first key is less than,
or greater than the second. equal to, or greater than the second
</entry> </entry>
<entry>1</entry> <entry>1</entry>
</row> </row>
<row> <row>
<entry>extractValue - extract keys from value to be indexed</entry> <entry>extractValue - extract keys from a value to be indexed</entry>
<entry>2</entry> <entry>2</entry>
</row> </row>
<row> <row>
<entry>extractQuery - extract keys from query</entry> <entry>extractQuery - extract keys from a query condition</entry>
<entry>3</entry> <entry>3</entry>
</row> </row>
<row> <row>
<entry>consistent - determine whether value matches by the <entry>consistent - determine whether value matches query condition</entry>
query</entry>
<entry>4</entry> <entry>4</entry>
</row> </row>
</tbody> </tbody>
@ -822,12 +821,16 @@ CREATE OPERATOR CLASS polygon_ops
STORAGE box; STORAGE box;
</programlisting> </programlisting>
At present, only the GiST and GIN index method supports a At present, only the GiST and GIN index methods support a
<literal>STORAGE</> type that's different from the column data type. <literal>STORAGE</> type that's different from the column data type.
The GiST <literal>compress</> and <literal>decompress</> support The GiST <function>compress</> and <function>decompress</> support
routines must deal with data-type conversion when <literal>STORAGE</> routines must deal with data-type conversion when <literal>STORAGE</>
is used. Functions named <literal>extractValue</> and <literal>extractQuery</> is used. In GIN, the <literal>STORAGE</> type identifies the type of
do conversation into internally used types for GIN. the <quote>key</> values, which normally is different from the type
of the indexed column &mdash; for example, an operator class for
integer array columns might have keys that are just integers. The
GIN <function>extractValue</> and <function>extractQuery</> support
routines are responsible for extracting keys from indexed values.
</para> </para>
</sect2> </sect2>