Move the "how to write a PL call handler" parts from the CREATE LANGUAGE

man page to the Programmer's Guide.
This commit is contained in:
Peter Eisentraut 2001-09-06 10:28:39 +00:00
parent f25ed23c57
commit 22ae53d4cd
2 changed files with 379 additions and 314 deletions

View File

@ -1,252 +1,196 @@
<!--
$Header: /cvsroot/pgsql/doc/src/sgml/ref/create_language.sgml,v 1.16 2001/09/03 12:57:49 petere Exp $
$Header: /cvsroot/pgsql/doc/src/sgml/ref/create_language.sgml,v 1.17 2001/09/06 10:28:39 petere Exp $
Postgres documentation
-->
<refentry id="SQL-CREATELANGUAGE">
<docinfo>
<date>2001-09-05</date>
</docinfo>
<refmeta>
<refentrytitle id="sql-createlanguage-title">
CREATE LANGUAGE
</refentrytitle>
<refentrytitle id="sql-createlanguage-title">CREATE LANGUAGE</refentrytitle>
<refmiscinfo>SQL - Language Statements</refmiscinfo>
</refmeta>
<refnamediv>
<refname>
CREATE LANGUAGE
</refname>
<refpurpose>
define a new procedural language
</refpurpose>
<refname>CREATE LANGUAGE</refname>
<refpurpose>define a new procedural language</refpurpose>
</refnamediv>
<refsynopsisdiv>
<refsynopsisdivinfo>
<date>1999-07-20</date>
</refsynopsisdivinfo>
<synopsis>
<synopsis>
CREATE [ TRUSTED ] [ PROCEDURAL ] LANGUAGE <replaceable class="parameter">langname</replaceable>
HANDLER <replaceable class="parameter">call_handler</replaceable>
</synopsis>
<refsect2 id="R2-SQL-CREATELANGUAGE-1">
<refsect2info>
<date>1998-09-09</date>
</refsect2info>
<title>
Inputs
</title>
<para>
<variablelist>
<varlistentry>
<term>TRUSTED</term>
<listitem>
<para>
<function> TRUSTED</function> specifies that the call handler for
the language is safe; that is, it offers an unprivileged user
no functionality to bypass access restrictions. If
this keyword is omitted when registering the language,
only users with the <productname>Postgres</productname>
superuser privilege can use
this language to create new functions.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term><replaceable class="parameter">langname</replaceable></term>
<listitem>
<para>
The name of the new procedural language.
The language name is case insensitive. A procedural
language cannot override one of the built-in languages of
<productname>Postgres</productname>.
</para>
<para>
For backward compatibility, the name may be enclosed by single
quotes.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term>HANDLER <replaceable class="parameter">call_handler</replaceable></term>
<listitem>
<para>
<replaceable class="parameter">call_handler</replaceable> is the name
of a previously
registered function that will be called to execute the PL
procedures.
</para>
</listitem>
</varlistentry>
</variablelist>
</para>
</refsect2>
<refsect2 id="R2-SQL-CREATELANGUAGE-2">
<refsect2info>
<date>1998-09-09</date>
</refsect2info>
<title>
Outputs
</title>
<para>
<variablelist>
<varlistentry>
<term><computeroutput>
CREATE
</computeroutput></term>
<listitem>
<para>
This message is returned if the language is successfully
created.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term><computeroutput>
ERROR: PL handler function <replaceable class="parameter">funcname</replaceable>() doesn't exist
</computeroutput></term>
<listitem>
<para>
This error is returned if the function
<replaceable class="parameter">funcname</replaceable>()
is not found.
</para>
</listitem>
</varlistentry>
</variablelist>
</para>
</refsect2>
</synopsis>
</refsynopsisdiv>
<refsect1 id="R1-SQL-CREATELANGUAGE-1">
<refsect1info>
<date>1998-09-09</date>
</refsect1info>
<title>
Description
</title>
<refsect1 id="sql-createlanguage-description">
<title>Description</title>
<para>
Using <command>CREATE LANGUAGE</command>, a
<productname>Postgres</productname> user can register
a new language with <productname>Postgres</productname>.
Subsequently, functions and
trigger procedures can be defined in this new language.
The user must have the <productname>Postgres</productname>
superuser privilege to
Using <command>CREATE LANGUAGE</command>, a
<productname>PostgreSQL</productname> user can register a new
procedural language with a <productname>PostgreSQL</productname>
database. Subsequently, functions and trigger procedures can be
defined in this new language. The user must have the
<productname>PostgreSQL</productname> superuser privilege to
register a new language.
</para>
<refsect2 id="R2-SQL-CREATELANGUAGE-3">
<refsect2info>
<date>1998-09-09</date>
</refsect2info>
<title>
Writing PL handlers
</title>
<para>
<command>CREATE LANGUAGE</command> effectively associates the
language name with a call handler that is responsible for executing
functions written in the language. Refer to the
<citetitle>Programmer's Guide</citetitle> for more information
about language call handlers.
</para>
<note>
<para>
Note that procedural languages are local to individual databases.
To make a language available in all databases by default, it should
be installed into the <literal>template1</literal> database.
</para>
</refsect1>
<refsect1 id="sql-createlanguage-parameters">
<title>Parameters</title>
<variablelist>
<varlistentry>
<term><literal>TRUSTED</literal></term>
<listitem>
<para>
<literal>TRUSTED</literal> specifies that the call handler for
the language is safe, that is, it does not offer an
unprivileged user any functionality to bypass access
restrictions. If this keyword is omitted when registering the
language, only users with the
<productname>PostgreSQL</productname> superuser privilege can
use this language to create new functions.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term><literal>PROCEDURAL</literal></term>
<listitem>
<para>
This is a noise word.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term><replaceable class="parameter">langname</replaceable></term>
<listitem>
<para>
The name of the new procedural language. The language name is
case insensitive. A procedural language cannot override one of
the built-in languages of <productname>PostgreSQL</productname>.
</para>
<para>
For backward compatibility, the name may be enclosed by single
quotes.
</para>
</listitem>
</varlistentry>
<varlistentry>
<term><literal>HANDLER</literal> <replaceable class="parameter">call_handler</replaceable></term>
<listitem>
<para>
<replaceable class="parameter">call_handler</replaceable> is
the name of a previously registered function that will be
called to execute the procedural language functions. The call
handler for a procedural language must be written in a compiled
language such as C with version 1 call convention and
registered with <productname>PostgreSQL</productname> as a
function taking no arguments and returning the
<type>opaque</type> type, a placeholder for unspecified or
undefined types.
</para>
</listitem>
</varlistentry>
</variablelist>
</refsect1>
<refsect1 id="sql-createlanguage-diagnostics">
<title>Diagnostics</title>
<msgset>
<msgentry>
<msg>
<msgmain>
<msgtext>
<screen>
CREATE
</screen>
</msgtext>
</msgmain>
</msg>
<msgexplan>
<para>
In <productname>Postgres</productname> 7.1 and later, call handlers
must adhere to the "version 1" function manager interface, not the
old-style interface.
This message is returned if the language is successfully
created.
</para>
</note>
</msgexplan>
</msgentry>
<para>
The call handler for a procedural language must be written
in a compiled language such as C and registered with
<productname>Postgres</productname> as a function taking
no arguments and returning the
<type>opaque</type> type, a placeholder for unspecified or undefined types.
This prevents the call handler from being
called directly as a function from queries.
(However, arguments may be supplied in the actual call when a
PL function in the language offered by the handler is to be executed.)
</para>
<msgentry>
<msg>
<msgmain>
<msgtext>
<screen>
ERROR: PL handler function <replaceable class="parameter">funcname</replaceable>() doesn't exist
</screen>
</msgtext>
</msgmain>
</msg>
<para>
The call handler is called in the same way as any other
function: it receives a pointer to a FunctionCallInfoData struct
containing argument values and information about the called function,
and it is expected to return a Datum result (and possibly set the
<literal>isnull</literal> field of the FunctionCallInfoData struct,
if it wishes to return an SQL NULL result). The difference between
a call handler and an ordinary callee function is that the
<literal>flinfo-&gt;fn_oid</literal> field of the FunctionCallInfoData
struct will contain the OID of the PL function to be called, not of
the call handler itself. The call handler must use this field to
determine which function to execute. Also, the passed argument list
has been set up according to the declaration of the target PL function,
not of the call handler.
</para>
<msgexplan>
<para>
This error is returned if the function <replaceable
class="parameter">funcname</replaceable>() is not found.
</para>
</msgexplan>
</msgentry>
</msgset>
</refsect1>
<para>
It's up to the call handler to fetch the
<filename>pg_proc</filename> entry and
to analyze the argument and return types of the called
procedure. The AS clause from the
<command>CREATE FUNCTION</command> of
the procedure will be found in the <literal>prosrc</literal>
attribute of the
<filename>pg_proc</filename> table entry. This may be the
source text in the procedural
language itself (like for PL/Tcl), a pathname to a
file, or anything else that tells the call handler what to
do in detail.
</para>
<refsect1 id="sql-createlanguage-notes">
<title>Notes</title>
<para>
Often, the same function is called many times per SQL statement.
A call handler can avoid repeated lookups of information about the
called function by using the <literal>flinfo-&gt;fn_extra</literal> field.
This will initially be NULL, but can be set by the call handler to
point at information about the PL function. On subsequent calls,
if <literal>flinfo-&gt;fn_extra</literal> is already non-NULL then it
can be used and the information lookup step skipped. The call handler
must be careful that <literal>flinfo-&gt;fn_extra</literal> is made to
point at memory that will live at least until the end of the current
query, since an FmgrInfo data structure could be kept that long.
One way to do this is to allocate the extra data in the memory context
specified by <literal>flinfo-&gt;fn_mcxt</literal>; such data will
normally have the same lifespan as the FmgrInfo itself. But the handler
could also choose to use a longer-lived context so that it can cache
function definition information across queries.
</para>
<para>
This command normally should not be executed directly by users.
For the procedural languages supplied in the
<productname>PostgreSQL</productname> distribution, the <xref
linkend="app-createlang"> program should be used, which will also
install the correct call handler. (<command>createlang</command>
will call <command>CREATE LANGUAGE</command> internally.)
</para>
<para>
When a PL function is invoked as a trigger, no explicit arguments
are passed, but the FunctionCallInfoData's
<literal>context</literal> field points at a TriggerData node,
rather than being NULL as it is in a plain function call.
A PL handler should provide mechanisms for PL functions to get
at the trigger information.
</para>
</refsect2>
<refsect2 id="R2-SQL-CREATELANGUAGE-4">
<refsect2info>
<date>1998-09-09</date>
</refsect2info>
<title>
Notes
</title>
<para>
Use <command>CREATE FUNCTION</command>
to create a function.
</para>
<para>
Use <command>DROP LANGUAGE</command> to drop procedural languages.
</para>
<para>
Refer to the table <filename>pg_language</filename>
for further information:
<programlisting>
<computeroutput>
<para>
Use the <xref linkend="sql-createfunction"> command to create a new
function.
</para>
<para>
Use <xref linkend="sql-droplanguage">, or better yet the <xref
linkend="app-droplang"> program, to drop procedural languages.
</para>
<para>
The system catalog <classname>pg_language</classname> records
information about the currently installed procedural languages.
<screen>
Table "pg_language"
Attribute | Type | Modifier
---------------+---------+----------
@ -261,107 +205,61 @@ ERROR: PL handler function <replaceable class="parameter">funcname</replaceable
internal | f | f | 0 | n/a
C | f | f | 0 | /bin/cc
sql | f | f | 0 | postgres
</computeroutput>
</programlisting>
</para>
</screen>
</para>
<para>
The call handler for a procedural language must normally be written
in C and registered as 'internal' or 'C' language, depending
on whether it is linked into the backend or dynamically loaded.
The call handler cannot use the old-style 'C' function interface.
</para>
<para>
At present, the definitions for a procedural language cannot be
changed once they have been created.
</para>
</refsect2>
<para>
At present, the definition of a procedural language cannot be
changed once is has been created.
</para>
</refsect1>
<refsect1 id="R1-SQL-CREATELANGUAGE-6">
<title>
Usage
</title>
<para>
This is a template for a PL handler written in C:
<programlisting>
#include "executor/spi.h"
#include "commands/trigger.h"
#include "utils/elog.h"
#include "fmgr.h"
#include "access/heapam.h"
#include "utils/syscache.h"
#include "catalog/pg_proc.h"
#include "catalog/pg_type.h"
PG_FUNCTION_INFO_V1(plsample_call_handler);
Datum
plsample_call_handler(PG_FUNCTION_ARGS)
{
Datum retval;
if (CALLED_AS_TRIGGER(fcinfo))
{
/*
* Called as a trigger procedure
*/
TriggerData *trigdata = (TriggerData *) fcinfo->context;
retval = ...
} else {
/*
* Called as a function
*/
retval = ...
}
return retval;
}
</programlisting>
</para>
<refsect1 id="sql-createlanguage-examples">
<title>Examples</title>
<para>
Only a few thousand lines of code have to be added instead
of the dots to complete the PL call handler.
See <command>CREATE FUNCTION</command> for information on how to compile
it into a loadable module.
</para>
<para>
The following commands then register the sample procedural
language:
<programlisting>
The following two commands executed in sequence will register a new
procedural language and the associated call handler.
<programlisting>
CREATE FUNCTION plsample_call_handler () RETURNS opaque
AS '/usr/local/pgsql/lib/plsample.so'
LANGUAGE C;
CREATE LANGUAGE plsample
HANDLER plsample_call_handler;
</programlisting>
</programlisting>
</para>
</refsect1>
<refsect1 id="R1-SQL-CREATELANGUAGE-7">
<title>
Compatibility
</title>
<refsect1 id="sql-createlanguage-compat">
<title>Compatibility</title>
<refsect2 id="R2-SQL-CREATELANGUAGE-5">
<refsect2info>
<date>1998-09-09</date>
</refsect2info>
<title>
SQL92
</title>
<para>
<command>CREATE LANGUAGE</command> is a
<productname>PostgreSQL</productname> extension.
</para>
</refsect1>
<para>
<command>CREATE LANGUAGE</command>
is a <productname>Postgres</productname> extension.
There is no <command>CREATE LANGUAGE</command> statement in
<acronym>SQL92</acronym>.
</para>
</refsect2>
<refsect1>
<title>History</title>
<para>
The <command>CREATE LANGUAGE</command> command first appeared in
<productname>PostgreSQL</productname> 6.3.
</para>
</refsect1>
<refsect1>
<title>See Also</title>
<para>
<simplelist type="inline">
<member><xref linkend="app-createlang"></member>
<member><xref linkend="sql-createfunction"></member>
<member><xref linkend="app-droplang"></member>
<member><xref linkend="sql-droplanguage"></member>
<member><citetitle>PostgreSQL Programmer's Guide</citetitle></member>
</simplelist>
</para>
</refsect1>
</refentry>

View File

@ -1,5 +1,5 @@
<!--
$Header: /cvsroot/pgsql/doc/src/sgml/xfunc.sgml,v 1.33 2001/08/28 14:20:26 petere Exp $
$Header: /cvsroot/pgsql/doc/src/sgml/xfunc.sgml,v 1.34 2001/09/06 10:28:39 petere Exp $
-->
<chapter id="xfunc">
@ -337,11 +337,11 @@ SELECT clean_EMP();
</para>
<para>
There are currently three procedural languages available in the standard
<productname>Postgres</productname> distribution (PLSQL, PLTCL and
PLPERL), and other languages can be defined.
Refer to <xref linkend="xplang"> for
more information.
There are currently four procedural languages available in the
standard <productname>PostgreSQL</productname> distribution:
PL/pgSQL, PL/Tcl, PL/Perl, and PL/Python. Other languages can be
defined by users. Refer to <xref linkend="xplang"> for more
information.
</para>
</sect1>
@ -1357,6 +1357,173 @@ LANGUAGE 'c';
</sect3>
</sect2>
</sect1>
<sect1 id="xfunc-plhandler">
<title>Procedural Language Handlers</title>
<para>
All calls to functions that are written in a language other than
the current <quote>version 1</quote> interface for compiled
languages, in particular in user-defined procedural languages, but
also functions written in SQL or the version 0 compiled language
interface, go through a <firstterm>call handler</firstterm>
function for the specific language. It is the responsibility of
the call handler to execute the function in a meaningful way, such
as by interpreting the supplied source text. This section
describes how a language call handler can be written. This is not
a common task, in fact, it has only been done a handful of times
in the history of <productname>PostgreSQL</productname>, but the
topic naturally belongs in this chapter, and the material might
give some insight into the extensible nature of the
<productname>PostgreSQL</productname> system.
</para>
<para>
The call handler for a procedural language is a
<quote>normal</quote> function, which must be written in a
compiled language such as C and registered with
<productname>PostgreSQL</productname> as taking no arguments and
returning the <type>opaque</type> type, a placeholder for
unspecified or undefined types. This prevents the call handler
from being called directly as a function from queries. (However,
arguments may be supplied in the actual call to the handler when a
function in the language offered by the handler is to be
executed.)
</para>
<note>
<para>
In <productname>PostgreSQL</productname> 7.1 and later, call
handlers must adhere to the <quote>version 1</quote> function
manager interface, not the old-style interface.
</para>
</note>
<para>
The call handler is called in the same way as any other function:
It receives a pointer to a
<structname>FunctionCallInfoData</structname> struct containing
argument values and information about the called function, and it
is expected to return a <type>Datum</type> result (and possibly
set the <structfield>isnull</structfield> field of the
<structname>FunctionCallInfoData</structname> struct, if it wishes
to return an SQL NULL result). The difference between a call
handler and an ordinary callee function is that the
<structfield>flinfo-&gt;fn_oid</structfield> field of the
<structname>FunctionCallInfoData</structname> struct will contain
the OID of the actual function to be called, not of the call
handler itself. The call handler must use this field to determine
which function to execute. Also, the passed argument list has
been set up according to the declaration of the target function,
not of the call handler.
</para>
<para>
It's up to the call handler to fetch the
<classname>pg_proc</classname> entry and to analyze the argument
and return types of the called procedure. The AS clause from the
<command>CREATE FUNCTION</command> of the procedure will be found
in the <literal>prosrc</literal> attribute of the
<classname>pg_proc</classname> table entry. This may be the source
text in the procedural language itself (like for PL/Tcl), a
pathname to a file, or anything else that tells the call handler
what to do in detail.
</para>
<para>
Often, the same function is called many times per SQL statement.
A call handler can avoid repeated lookups of information about the
called function by using the
<structfield>flinfo-&gt;fn_extra</structfield> field. This will
initially be NULL, but can be set by the call handler to point at
information about the PL function. On subsequent calls, if
<structfield>flinfo-&gt;fn_extra</structfield> is already non-NULL
then it can be used and the information lookup step skipped. The
call handler must be careful that
<structfield>flinfo-&gt;fn_extra</structfield> is made to point at
memory that will live at least until the end of the current query,
since an <structname>FmgrInfo</structname> data structure could be
kept that long. One way to do this is to allocate the extra data
in the memory context specified by
<structfield>flinfo-&gt;fn_mcxt</structfield>; such data will
normally have the same lifespan as the
<structname>FmgrInfo</structname> itself. But the handler could
also choose to use a longer-lived context so that it can cache
function definition information across queries.
</para>
<para>
When a PL function is invoked as a trigger, no explicit arguments
are passed, but the
<structname>FunctionCallInfoData</structname>'s
<structfield>context</structfield> field points at a
<structname>TriggerData</structname> node, rather than being NULL
as it is in a plain function call. A language handler should
provide mechanisms for PL functions to get at the trigger
information.
</para>
<para>
This is a template for a PL handler written in C:
<programlisting>
#include "postgres.h"
#include "executor/spi.h"
#include "commands/trigger.h"
#include "utils/elog.h"
#include "fmgr.h"
#include "access/heapam.h"
#include "utils/syscache.h"
#include "catalog/pg_proc.h"
#include "catalog/pg_type.h"
PG_FUNCTION_INFO_V1(plsample_call_handler);
Datum
plsample_call_handler(PG_FUNCTION_ARGS)
{
Datum retval;
if (CALLED_AS_TRIGGER(fcinfo))
{
/*
* Called as a trigger procedure
*/
TriggerData *trigdata = (TriggerData *) fcinfo->context;
retval = ...
}
else {
/*
* Called as a function
*/
retval = ...
}
return retval;
}
</programlisting>
</para>
<para>
Only a few thousand lines of code have to be added instead of the
dots to complete the call handler. See <xref linkend="xfunc-c">
for information on how to compile it into a loadable module.
</para>
<para>
The following commands then register the sample procedural
language:
<programlisting>
CREATE FUNCTION plsample_call_handler () RETURNS opaque
AS '/usr/local/pgsql/lib/plsample.so'
LANGUAGE C;
CREATE LANGUAGE plsample
HANDLER plsample_call_handler;
</programlisting>
</para>
</sect1>
</chapter>
<!-- Keep this comment at the end of the file