ok
Direktori : /opt/alt/postgresql11/usr/share/doc/alt-postgresql11-9.2.24/html/ |
Current File : //opt/alt/postgresql11/usr/share/doc/alt-postgresql11-9.2.24/html/xfunc-c.html |
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd"> <HTML ><HEAD ><TITLE >C-Language Functions</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="Extending SQL" HREF="extend.html"><LINK REL="PREVIOUS" TITLE="Internal Functions" HREF="xfunc-internal.html"><LINK REL="NEXT" TITLE="User-defined Aggregates" HREF="xaggr.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="Internal Functions" HREF="xfunc-internal.html" ACCESSKEY="P" >Prev</A ></TD ><TD WIDTH="10%" ALIGN="left" VALIGN="top" ><A HREF="extend.html" ACCESSKEY="U" >Up</A ></TD ><TD WIDTH="60%" ALIGN="center" VALIGN="bottom" >Chapter 35. Extending <ACRONYM CLASS="ACRONYM" >SQL</ACRONYM ></TD ><TD WIDTH="20%" ALIGN="right" VALIGN="top" ><A TITLE="User-defined Aggregates" HREF="xaggr.html" ACCESSKEY="N" >Next</A ></TD ></TR ></TABLE ><HR ALIGN="LEFT" WIDTH="100%"></DIV ><DIV CLASS="SECT1" ><H1 CLASS="SECT1" ><A NAME="XFUNC-C" >35.9. C-Language Functions</A ></H1 ><P > User-defined functions can be written in C (or a language that can be made compatible with C, such as C++). Such functions are compiled into dynamically loadable objects (also called shared libraries) and are loaded by the server on demand. The dynamic loading feature is what distinguishes <SPAN CLASS="QUOTE" >"C language"</SPAN > functions from <SPAN CLASS="QUOTE" >"internal"</SPAN > functions — the actual coding conventions are essentially the same for both. (Hence, the standard internal function library is a rich source of coding examples for user-defined C functions.) </P ><P > Two different calling conventions are currently used for C functions. The newer <SPAN CLASS="QUOTE" >"version 1"</SPAN > calling convention is indicated by writing a <TT CLASS="LITERAL" >PG_FUNCTION_INFO_V1()</TT > macro call for the function, as illustrated below. Lack of such a macro indicates an old-style (<SPAN CLASS="QUOTE" >"version 0"</SPAN >) function. The language name specified in <TT CLASS="COMMAND" >CREATE FUNCTION</TT > is <TT CLASS="LITERAL" >C</TT > in either case. Old-style functions are now deprecated because of portability problems and lack of functionality, but they are still supported for compatibility reasons. </P ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="XFUNC-C-DYNLOAD" >35.9.1. Dynamic Loading</A ></H2 ><P > The first time a user-defined function in a particular loadable object file is called in a session, the dynamic loader loads that object file into memory so that the function can be called. The <TT CLASS="COMMAND" >CREATE FUNCTION</TT > for a user-defined C function must therefore specify two pieces of information for the function: the name of the loadable object file, and the C name (link symbol) of the specific function to call within that object file. If the C name is not explicitly specified then it is assumed to be the same as the SQL function name. </P ><P > The following algorithm is used to locate the shared object file based on the name given in the <TT CLASS="COMMAND" >CREATE FUNCTION</TT > command: <P ></P ></P><OL TYPE="1" ><LI ><P > If the name is an absolute path, the given file is loaded. </P ></LI ><LI ><P > If the name starts with the string <TT CLASS="LITERAL" >$libdir</TT >, that part is replaced by the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > package library directory name, which is determined at build time. </P ></LI ><LI ><P > If the name does not contain a directory part, the file is searched for in the path specified by the configuration variable <A HREF="runtime-config-client.html#GUC-DYNAMIC-LIBRARY-PATH" >dynamic_library_path</A >. </P ></LI ><LI ><P > Otherwise (the file was not found in the path, or it contains a non-absolute directory part), the dynamic loader will try to take the name as given, which will most likely fail. (It is unreliable to depend on the current working directory.) </P ></LI ></OL ><P> If this sequence does not work, the platform-specific shared library file name extension (often <TT CLASS="FILENAME" >.so</TT >) is appended to the given name and this sequence is tried again. If that fails as well, the load will fail. </P ><P > It is recommended to locate shared libraries either relative to <TT CLASS="LITERAL" >$libdir</TT > or through the dynamic library path. This simplifies version upgrades if the new installation is at a different location. The actual directory that <TT CLASS="LITERAL" >$libdir</TT > stands for can be found out with the command <TT CLASS="LITERAL" >pg_config --pkglibdir</TT >. </P ><P > The user ID the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > server runs as must be able to traverse the path to the file you intend to load. Making the file or a higher-level directory not readable and/or not executable by the <SPAN CLASS="SYSTEMITEM" >postgres</SPAN > user is a common mistake. </P ><P > In any case, the file name that is given in the <TT CLASS="COMMAND" >CREATE FUNCTION</TT > command is recorded literally in the system catalogs, so if the file needs to be loaded again the same procedure is applied. </P ><DIV CLASS="NOTE" ><BLOCKQUOTE CLASS="NOTE" ><P ><B >Note: </B > <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > will not compile a C function automatically. The object file must be compiled before it is referenced in a <TT CLASS="COMMAND" >CREATE FUNCTION</TT > command. See <A HREF="xfunc-c.html#DFUNC" >Section 35.9.6</A > for additional information. </P ></BLOCKQUOTE ></DIV ><P > To ensure that a dynamically loaded object file is not loaded into an incompatible server, <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > checks that the file contains a <SPAN CLASS="QUOTE" >"magic block"</SPAN > with the appropriate contents. This allows the server to detect obvious incompatibilities, such as code compiled for a different major version of <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN >. A magic block is required as of <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > 8.2. To include a magic block, write this in one (and only one) of the module source files, after having included the header <TT CLASS="FILENAME" >fmgr.h</TT >: </P><PRE CLASS="PROGRAMLISTING" >#ifdef PG_MODULE_MAGIC PG_MODULE_MAGIC; #endif</PRE ><P> The <TT CLASS="LITERAL" >#ifdef</TT > test can be omitted if the code doesn't need to compile against pre-8.2 <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > releases. </P ><P > After it is used for the first time, a dynamically loaded object file is retained in memory. Future calls in the same session to the function(s) in that file will only incur the small overhead of a symbol table lookup. If you need to force a reload of an object file, for example after recompiling it, begin a fresh session. </P ><P > Optionally, a dynamically loaded file can contain initialization and finalization functions. If the file includes a function named <CODE CLASS="FUNCTION" >_PG_init</CODE >, that function will be called immediately after loading the file. The function receives no parameters and should return void. If the file includes a function named <CODE CLASS="FUNCTION" >_PG_fini</CODE >, that function will be called immediately before unloading the file. Likewise, the function receives no parameters and should return void. Note that <CODE CLASS="FUNCTION" >_PG_fini</CODE > will only be called during an unload of the file, not during process termination. (Presently, unloads are disabled and will never occur, but this may change in the future.) </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="XFUNC-C-BASETYPE" >35.9.2. Base Types in C-Language Functions</A ></H2 ><P > To know how to write C-language functions, you need to know how <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > internally represents base data types and how they can be passed to and from functions. Internally, <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > regards a base type as a <SPAN CLASS="QUOTE" >"blob of memory"</SPAN >. The user-defined functions that you define over a type in turn define the way that <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > can operate on it. That is, <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > will only store and retrieve the data from disk and use your user-defined functions to input, process, and output the data. </P ><P > Base types can have one of three internal formats: <P ></P ></P><UL ><LI ><P > pass by value, fixed-length </P ></LI ><LI ><P > pass by reference, fixed-length </P ></LI ><LI ><P > pass by reference, variable-length </P ></LI ></UL ><P> </P ><P > By-value types can only be 1, 2, or 4 bytes in length (also 8 bytes, if <TT CLASS="LITERAL" >sizeof(Datum)</TT > is 8 on your machine). You should be careful to define your types such that they will be the same size (in bytes) on all architectures. For example, the <TT CLASS="LITERAL" >long</TT > type is dangerous because it is 4 bytes on some machines and 8 bytes on others, whereas <TT CLASS="TYPE" >int</TT > type is 4 bytes on most Unix machines. A reasonable implementation of the <TT CLASS="TYPE" >int4</TT > type on Unix machines might be: </P><PRE CLASS="PROGRAMLISTING" >/* 4-byte integer, passed by value */ typedef int int4;</PRE ><P> </P ><P > On the other hand, fixed-length types of any size can be passed by-reference. For example, here is a sample implementation of a <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > type: </P><PRE CLASS="PROGRAMLISTING" >/* 16-byte structure, passed by reference */ typedef struct { double x, y; } Point;</PRE ><P> Only pointers to such types can be used when passing them in and out of <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > functions. To return a value of such a type, allocate the right amount of memory with <TT CLASS="LITERAL" >palloc</TT >, fill in the allocated memory, and return a pointer to it. (Also, if you just want to return the same value as one of your input arguments that's of the same data type, you can skip the extra <TT CLASS="LITERAL" >palloc</TT > and just return the pointer to the input value.) </P ><P > Finally, all variable-length types must also be passed by reference. All variable-length types must begin with an opaque length field of exactly 4 bytes, which will be set by <TT CLASS="SYMBOL" >SET_VARSIZE</TT >; never set this field directly! All data to be stored within that type must be located in the memory immediately following that length field. The length field contains the total length of the structure, that is, it includes the size of the length field itself. </P ><P > Another important point is to avoid leaving any uninitialized bits within data type values; for example, take care to zero out any alignment padding bytes that might be present in structs. Without this, logically-equivalent constants of your data type might be seen as unequal by the planner, leading to inefficient (though not incorrect) plans. </P ><DIV CLASS="WARNING" ><P ></P ><TABLE CLASS="WARNING" BORDER="1" WIDTH="100%" ><TR ><TD ALIGN="CENTER" ><B >Warning</B ></TD ></TR ><TR ><TD ALIGN="LEFT" ><P > <SPAN CLASS="emphasis" ><I CLASS="EMPHASIS" >Never</I ></SPAN > modify the contents of a pass-by-reference input value. If you do so you are likely to corrupt on-disk data, since the pointer you are given might point directly into a disk buffer. The sole exception to this rule is explained in <A HREF="xaggr.html" >Section 35.10</A >. </P ></TD ></TR ></TABLE ></DIV ><P > As an example, we can define the type <TT CLASS="TYPE" >text</TT > as follows: </P><PRE CLASS="PROGRAMLISTING" >typedef struct { int4 length; char data[1]; } text;</PRE ><P> Obviously, the data field declared here is not long enough to hold all possible strings. Since it's impossible to declare a variable-size structure in <ACRONYM CLASS="ACRONYM" >C</ACRONYM >, we rely on the knowledge that the <ACRONYM CLASS="ACRONYM" >C</ACRONYM > compiler won't range-check array subscripts. We just allocate the necessary amount of space and then access the array as if it were declared the right length. (This is a common trick, which you can read about in many textbooks about C.) </P ><P > When manipulating variable-length types, we must be careful to allocate the correct amount of memory and set the length field correctly. For example, if we wanted to store 40 bytes in a <TT CLASS="STRUCTNAME" >text</TT > structure, we might use a code fragment like this: </P><PRE CLASS="PROGRAMLISTING" >#include "postgres.h" ... char buffer[40]; /* our source data */ ... text *destination = (text *) palloc(VARHDRSZ + 40); SET_VARSIZE(destination, VARHDRSZ + 40); memcpy(destination->data, buffer, 40); ...</PRE ><P> <TT CLASS="LITERAL" >VARHDRSZ</TT > is the same as <TT CLASS="LITERAL" >sizeof(int32)</TT >, but it's considered good style to use the macro <TT CLASS="LITERAL" >VARHDRSZ</TT > to refer to the size of the overhead for a variable-length type. Also, the length field <SPAN CLASS="emphasis" ><I CLASS="EMPHASIS" >must</I ></SPAN > be set using the <TT CLASS="LITERAL" >SET_VARSIZE</TT > macro, not by simple assignment. </P ><P > <A HREF="xfunc-c.html#XFUNC-C-TYPE-TABLE" >Table 35-1</A > specifies which C type corresponds to which SQL type when writing a C-language function that uses a built-in type of <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN >. The <SPAN CLASS="QUOTE" >"Defined In"</SPAN > column gives the header file that needs to be included to get the type definition. (The actual definition might be in a different file that is included by the listed file. It is recommended that users stick to the defined interface.) Note that you should always include <TT CLASS="FILENAME" >postgres.h</TT > first in any source file, because it declares a number of things that you will need anyway. </P ><DIV CLASS="TABLE" ><A NAME="XFUNC-C-TYPE-TABLE" ></A ><P ><B >Table 35-1. Equivalent C Types for Built-in SQL Types</B ></P ><TABLE BORDER="1" CLASS="CALSTABLE" ><COL><COL><COL><THEAD ><TR ><TH > SQL Type </TH ><TH > C Type </TH ><TH > Defined In </TH ></TR ></THEAD ><TBODY ><TR ><TD ><TT CLASS="TYPE" >abstime</TT ></TD ><TD ><TT CLASS="TYPE" >AbsoluteTime</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/nabstime.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >boolean</TT ></TD ><TD ><TT CLASS="TYPE" >bool</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT > (maybe compiler built-in)</TD ></TR ><TR ><TD ><TT CLASS="TYPE" >box</TT ></TD ><TD ><TT CLASS="TYPE" >BOX*</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/geo_decls.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >bytea</TT ></TD ><TD ><TT CLASS="TYPE" >bytea*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >"char"</TT ></TD ><TD ><TT CLASS="TYPE" >char</TT ></TD ><TD >(compiler built-in)</TD ></TR ><TR ><TD ><TT CLASS="TYPE" >character</TT ></TD ><TD ><TT CLASS="TYPE" >BpChar*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >cid</TT ></TD ><TD ><TT CLASS="TYPE" >CommandId</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >date</TT ></TD ><TD ><TT CLASS="TYPE" >DateADT</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/date.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >smallint</TT > (<TT CLASS="TYPE" >int2</TT >)</TD ><TD ><TT CLASS="TYPE" >int2</TT > or <TT CLASS="TYPE" >int16</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >int2vector</TT ></TD ><TD ><TT CLASS="TYPE" >int2vector*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >integer</TT > (<TT CLASS="TYPE" >int4</TT >)</TD ><TD ><TT CLASS="TYPE" >int4</TT > or <TT CLASS="TYPE" >int32</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >real</TT > (<TT CLASS="TYPE" >float4</TT >)</TD ><TD ><TT CLASS="TYPE" >float4*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >double precision</TT > (<TT CLASS="TYPE" >float8</TT >)</TD ><TD ><TT CLASS="TYPE" >float8*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >interval</TT ></TD ><TD ><TT CLASS="TYPE" >Interval*</TT ></TD ><TD ><TT CLASS="FILENAME" >datatype/timestamp.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >lseg</TT ></TD ><TD ><TT CLASS="TYPE" >LSEG*</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/geo_decls.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >name</TT ></TD ><TD ><TT CLASS="TYPE" >Name</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >oid</TT ></TD ><TD ><TT CLASS="TYPE" >Oid</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >oidvector</TT ></TD ><TD ><TT CLASS="TYPE" >oidvector*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >path</TT ></TD ><TD ><TT CLASS="TYPE" >PATH*</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/geo_decls.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >point</TT ></TD ><TD ><TT CLASS="TYPE" >POINT*</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/geo_decls.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >regproc</TT ></TD ><TD ><TT CLASS="TYPE" >regproc</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >reltime</TT ></TD ><TD ><TT CLASS="TYPE" >RelativeTime</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/nabstime.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >text</TT ></TD ><TD ><TT CLASS="TYPE" >text*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >tid</TT ></TD ><TD ><TT CLASS="TYPE" >ItemPointer</TT ></TD ><TD ><TT CLASS="FILENAME" >storage/itemptr.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >time</TT ></TD ><TD ><TT CLASS="TYPE" >TimeADT</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/date.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >time with time zone</TT ></TD ><TD ><TT CLASS="TYPE" >TimeTzADT</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/date.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >timestamp</TT ></TD ><TD ><TT CLASS="TYPE" >Timestamp*</TT ></TD ><TD ><TT CLASS="FILENAME" >datatype/timestamp.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >tinterval</TT ></TD ><TD ><TT CLASS="TYPE" >TimeInterval</TT ></TD ><TD ><TT CLASS="FILENAME" >utils/nabstime.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >varchar</TT ></TD ><TD ><TT CLASS="TYPE" >VarChar*</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ><TR ><TD ><TT CLASS="TYPE" >xid</TT ></TD ><TD ><TT CLASS="TYPE" >TransactionId</TT ></TD ><TD ><TT CLASS="FILENAME" >postgres.h</TT ></TD ></TR ></TBODY ></TABLE ></DIV ><P > Now that we've gone over all of the possible structures for base types, we can show some examples of real functions. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="AEN53110" >35.9.3. Version 0 Calling Conventions</A ></H2 ><P > We present the <SPAN CLASS="QUOTE" >"old style"</SPAN > calling convention first — although this approach is now deprecated, it's easier to get a handle on initially. In the version-0 method, the arguments and result of the C function are just declared in normal C style, but being careful to use the C representation of each SQL data type as shown above. </P ><P > Here are some examples: </P><PRE CLASS="PROGRAMLISTING" >#include "postgres.h" #include <string.h> #include "utils/geo_decls.h" #ifdef PG_MODULE_MAGIC PG_MODULE_MAGIC; #endif /* by value */ int add_one(int arg) { return arg + 1; } /* by reference, fixed length */ float8 * add_one_float8(float8 *arg) { float8 *result = (float8 *) palloc(sizeof(float8)); *result = *arg + 1.0; return result; } Point * makepoint(Point *pointx, Point *pointy) { Point *new_point = (Point *) palloc(sizeof(Point)); new_point->x = pointx->x; new_point->y = pointy->y; return new_point; } /* by reference, variable length */ text * copytext(text *t) { /* * VARSIZE is the total size of the struct in bytes. */ text *new_t = (text *) palloc(VARSIZE(t)); SET_VARSIZE(new_t, VARSIZE(t)); /* * VARDATA is a pointer to the data region of the struct. */ memcpy((void *) VARDATA(new_t), /* destination */ (void *) VARDATA(t), /* source */ VARSIZE(t) - VARHDRSZ); /* how many bytes */ return new_t; } text * concat_text(text *arg1, text *arg2) { int32 new_text_size = VARSIZE(arg1) + VARSIZE(arg2) - VARHDRSZ; text *new_text = (text *) palloc(new_text_size); SET_VARSIZE(new_text, new_text_size); memcpy(VARDATA(new_text), VARDATA(arg1), VARSIZE(arg1) - VARHDRSZ); memcpy(VARDATA(new_text) + (VARSIZE(arg1) - VARHDRSZ), VARDATA(arg2), VARSIZE(arg2) - VARHDRSZ); return new_text; }</PRE ><P> </P ><P > Supposing that the above code has been prepared in file <TT CLASS="FILENAME" >funcs.c</TT > and compiled into a shared object, we could define the functions to <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > with commands like this: </P><PRE CLASS="PROGRAMLISTING" >CREATE FUNCTION add_one(integer) RETURNS integer AS '<TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT >/funcs', 'add_one' LANGUAGE C STRICT; -- note overloading of SQL function name "add_one" CREATE FUNCTION add_one(double precision) RETURNS double precision AS '<TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT >/funcs', 'add_one_float8' LANGUAGE C STRICT; CREATE FUNCTION makepoint(point, point) RETURNS point AS '<TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT >/funcs', 'makepoint' LANGUAGE C STRICT; CREATE FUNCTION copytext(text) RETURNS text AS '<TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT >/funcs', 'copytext' LANGUAGE C STRICT; CREATE FUNCTION concat_text(text, text) RETURNS text AS '<TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT >/funcs', 'concat_text' LANGUAGE C STRICT;</PRE ><P> </P ><P > Here, <TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT > stands for the directory of the shared library file (for instance the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > tutorial directory, which contains the code for the examples used in this section). (Better style would be to use just <TT CLASS="LITERAL" >'funcs'</TT > in the <TT CLASS="LITERAL" >AS</TT > clause, after having added <TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT > to the search path. In any case, we can omit the system-specific extension for a shared library, commonly <TT CLASS="LITERAL" >.so</TT > or <TT CLASS="LITERAL" >.sl</TT >.) </P ><P > Notice that we have specified the functions as <SPAN CLASS="QUOTE" >"strict"</SPAN >, meaning that the system should automatically assume a null result if any input value is null. By doing this, we avoid having to check for null inputs in the function code. Without this, we'd have to check for null values explicitly, by checking for a null pointer for each pass-by-reference argument. (For pass-by-value arguments, we don't even have a way to check!) </P ><P > Although this calling convention is simple to use, it is not very portable; on some architectures there are problems with passing data types that are smaller than <TT CLASS="TYPE" >int</TT > this way. Also, there is no simple way to return a null result, nor to cope with null arguments in any way other than making the function strict. The version-1 convention, presented next, overcomes these objections. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="AEN53137" >35.9.4. Version 1 Calling Conventions</A ></H2 ><P > The version-1 calling convention relies on macros to suppress most of the complexity of passing arguments and results. The C declaration of a version-1 function is always: </P><PRE CLASS="PROGRAMLISTING" >Datum funcname(PG_FUNCTION_ARGS)</PRE ><P> In addition, the macro call: </P><PRE CLASS="PROGRAMLISTING" >PG_FUNCTION_INFO_V1(funcname);</PRE ><P> must appear in the same source file. (Conventionally, it's written just before the function itself.) This macro call is not needed for <TT CLASS="LITERAL" >internal</TT >-language functions, since <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > assumes that all internal functions use the version-1 convention. It is, however, required for dynamically-loaded functions. </P ><P > In a version-1 function, each actual argument is fetched using a <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >()</CODE > macro that corresponds to the argument's data type, and the result is returned using a <CODE CLASS="FUNCTION" >PG_RETURN_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >()</CODE > macro for the return type. <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >()</CODE > takes as its argument the number of the function argument to fetch, where the count starts at 0. <CODE CLASS="FUNCTION" >PG_RETURN_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >()</CODE > takes as its argument the actual value to return. </P ><P > Here we show the same functions as above, coded in version-1 style: </P><PRE CLASS="PROGRAMLISTING" >#include "postgres.h" #include <string.h> #include "fmgr.h" #include "utils/geo_decls.h" #ifdef PG_MODULE_MAGIC PG_MODULE_MAGIC; #endif /* by value */ PG_FUNCTION_INFO_V1(add_one); Datum add_one(PG_FUNCTION_ARGS) { int32 arg = PG_GETARG_INT32(0); PG_RETURN_INT32(arg + 1); } /* by reference, fixed length */ PG_FUNCTION_INFO_V1(add_one_float8); Datum add_one_float8(PG_FUNCTION_ARGS) { /* The macros for FLOAT8 hide its pass-by-reference nature. */ float8 arg = PG_GETARG_FLOAT8(0); PG_RETURN_FLOAT8(arg + 1.0); } PG_FUNCTION_INFO_V1(makepoint); Datum makepoint(PG_FUNCTION_ARGS) { /* Here, the pass-by-reference nature of Point is not hidden. */ Point *pointx = PG_GETARG_POINT_P(0); Point *pointy = PG_GETARG_POINT_P(1); Point *new_point = (Point *) palloc(sizeof(Point)); new_point->x = pointx->x; new_point->y = pointy->y; PG_RETURN_POINT_P(new_point); } /* by reference, variable length */ PG_FUNCTION_INFO_V1(copytext); Datum copytext(PG_FUNCTION_ARGS) { text *t = PG_GETARG_TEXT_P(0); /* * VARSIZE is the total size of the struct in bytes. */ text *new_t = (text *) palloc(VARSIZE(t)); SET_VARSIZE(new_t, VARSIZE(t)); /* * VARDATA is a pointer to the data region of the struct. */ memcpy((void *) VARDATA(new_t), /* destination */ (void *) VARDATA(t), /* source */ VARSIZE(t) - VARHDRSZ); /* how many bytes */ PG_RETURN_TEXT_P(new_t); } PG_FUNCTION_INFO_V1(concat_text); Datum concat_text(PG_FUNCTION_ARGS) { text *arg1 = PG_GETARG_TEXT_P(0); text *arg2 = PG_GETARG_TEXT_P(1); int32 new_text_size = VARSIZE(arg1) + VARSIZE(arg2) - VARHDRSZ; text *new_text = (text *) palloc(new_text_size); SET_VARSIZE(new_text, new_text_size); memcpy(VARDATA(new_text), VARDATA(arg1), VARSIZE(arg1) - VARHDRSZ); memcpy(VARDATA(new_text) + (VARSIZE(arg1) - VARHDRSZ), VARDATA(arg2), VARSIZE(arg2) - VARHDRSZ); PG_RETURN_TEXT_P(new_text); }</PRE ><P> </P ><P > The <TT CLASS="COMMAND" >CREATE FUNCTION</TT > commands are the same as for the version-0 equivalents. </P ><P > At first glance, the version-1 coding conventions might appear to be just pointless obscurantism. They do, however, offer a number of improvements, because the macros can hide unnecessary detail. An example is that in coding <CODE CLASS="FUNCTION" >add_one_float8</CODE >, we no longer need to be aware that <TT CLASS="TYPE" >float8</TT > is a pass-by-reference type. Another example is that the <TT CLASS="LITERAL" >GETARG</TT > macros for variable-length types allow for more efficient fetching of <SPAN CLASS="QUOTE" >"toasted"</SPAN > (compressed or out-of-line) values. </P ><P > One big improvement in version-1 functions is better handling of null inputs and results. The macro <CODE CLASS="FUNCTION" >PG_ARGISNULL(<TT CLASS="REPLACEABLE" ><I >n</I ></TT >)</CODE > allows a function to test whether each input is null. (Of course, doing this is only necessary in functions not declared <SPAN CLASS="QUOTE" >"strict"</SPAN >.) As with the <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >()</CODE > macros, the input arguments are counted beginning at zero. Note that one should refrain from executing <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >()</CODE > until one has verified that the argument isn't null. To return a null result, execute <CODE CLASS="FUNCTION" >PG_RETURN_NULL()</CODE >; this works in both strict and nonstrict functions. </P ><P > Other options provided in the new-style interface are two variants of the <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >()</CODE > macros. The first of these, <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >_COPY()</CODE >, guarantees to return a copy of the specified argument that is safe for writing into. (The normal macros will sometimes return a pointer to a value that is physically stored in a table, which must not be written to. Using the <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >_COPY()</CODE > macros guarantees a writable result.) The second variant consists of the <CODE CLASS="FUNCTION" >PG_GETARG_<TT CLASS="REPLACEABLE" ><I >xxx</I ></TT >_SLICE()</CODE > macros which take three arguments. The first is the number of the function argument (as above). The second and third are the offset and length of the segment to be returned. Offsets are counted from zero, and a negative length requests that the remainder of the value be returned. These macros provide more efficient access to parts of large values in the case where they have storage type <SPAN CLASS="QUOTE" >"external"</SPAN >. (The storage type of a column can be specified using <TT CLASS="LITERAL" >ALTER TABLE <TT CLASS="REPLACEABLE" ><I >tablename</I ></TT > ALTER COLUMN <TT CLASS="REPLACEABLE" ><I >colname</I ></TT > SET STORAGE <TT CLASS="REPLACEABLE" ><I >storagetype</I ></TT ></TT >. <TT CLASS="REPLACEABLE" ><I >storagetype</I ></TT > is one of <TT CLASS="LITERAL" >plain</TT >, <TT CLASS="LITERAL" >external</TT >, <TT CLASS="LITERAL" >extended</TT >, or <TT CLASS="LITERAL" >main</TT >.) </P ><P > Finally, the version-1 function call conventions make it possible to return set results (<A HREF="xfunc-c.html#XFUNC-C-RETURN-SET" >Section 35.9.9</A >) and implement trigger functions (<A HREF="triggers.html" >Chapter 36</A >) and procedural-language call handlers (<A HREF="plhandler.html" >Chapter 49</A >). Version-1 code is also more portable than version-0, because it does not break restrictions on function call protocol in the C standard. For more details see <TT CLASS="FILENAME" >src/backend/utils/fmgr/README</TT > in the source distribution. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="AEN53195" >35.9.5. Writing Code</A ></H2 ><P > Before we turn to the more advanced topics, we should discuss some coding rules for <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > C-language functions. While it might be possible to load functions written in languages other than C into <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN >, this is usually difficult (when it is possible at all) because other languages, such as C++, FORTRAN, or Pascal often do not follow the same calling convention as C. That is, other languages do not pass argument and return values between functions in the same way. For this reason, we will assume that your C-language functions are actually written in C. </P ><P > The basic rules for writing and building C functions are as follows: <P ></P ></P><UL ><LI ><P > Use <TT CLASS="LITERAL" >pg_config --includedir-server</TT > to find out where the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > server header files are installed on your system (or the system that your users will be running on). </P ></LI ><LI ><P > Compiling and linking your code so that it can be dynamically loaded into <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > always requires special flags. See <A HREF="xfunc-c.html#DFUNC" >Section 35.9.6</A > for a detailed explanation of how to do it for your particular operating system. </P ></LI ><LI ><P > Remember to define a <SPAN CLASS="QUOTE" >"magic block"</SPAN > for your shared library, as described in <A HREF="xfunc-c.html#XFUNC-C-DYNLOAD" >Section 35.9.1</A >. </P ></LI ><LI ><P > When allocating memory, use the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > functions <CODE CLASS="FUNCTION" >palloc</CODE > and <CODE CLASS="FUNCTION" >pfree</CODE > instead of the corresponding C library functions <CODE CLASS="FUNCTION" >malloc</CODE > and <CODE CLASS="FUNCTION" >free</CODE >. The memory allocated by <CODE CLASS="FUNCTION" >palloc</CODE > will be freed automatically at the end of each transaction, preventing memory leaks. </P ></LI ><LI ><P > Always zero the bytes of your structures using <CODE CLASS="FUNCTION" >memset</CODE > (or allocate them with <CODE CLASS="FUNCTION" >palloc0</CODE > in the first place). Even if you assign to each field of your structure, there might be alignment padding (holes in the structure) that contain garbage values. Without this, it's difficult to support hash indexes or hash joins, as you must pick out only the significant bits of your data structure to compute a hash. The planner also sometimes relies on comparing constants via bitwise equality, so you can get undesirable planning results if logically-equivalent values aren't bitwise equal. </P ></LI ><LI ><P > Most of the internal <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > types are declared in <TT CLASS="FILENAME" >postgres.h</TT >, while the function manager interfaces (<TT CLASS="SYMBOL" >PG_FUNCTION_ARGS</TT >, etc.) are in <TT CLASS="FILENAME" >fmgr.h</TT >, so you will need to include at least these two files. For portability reasons it's best to include <TT CLASS="FILENAME" >postgres.h</TT > <SPAN CLASS="emphasis" ><I CLASS="EMPHASIS" >first</I ></SPAN >, before any other system or user header files. Including <TT CLASS="FILENAME" >postgres.h</TT > will also include <TT CLASS="FILENAME" >elog.h</TT > and <TT CLASS="FILENAME" >palloc.h</TT > for you. </P ></LI ><LI ><P > Symbol names defined within object files must not conflict with each other or with symbols defined in the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > server executable. You will have to rename your functions or variables if you get error messages to this effect. </P ></LI ></UL ><P> </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="DFUNC" >35.9.6. Compiling and Linking Dynamically-loaded Functions</A ></H2 ><P > Before you are able to use your <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > extension functions written in C, they must be compiled and linked in a special way to produce a file that can be dynamically loaded by the server. To be precise, a <I CLASS="FIRSTTERM" >shared library</I > needs to be created. </P ><P > For information beyond what is contained in this section you should read the documentation of your operating system, in particular the manual pages for the C compiler, <TT CLASS="COMMAND" >cc</TT >, and the link editor, <TT CLASS="COMMAND" >ld</TT >. In addition, the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > source code contains several working examples in the <TT CLASS="FILENAME" >contrib</TT > directory. If you rely on these examples you will make your modules dependent on the availability of the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > source code, however. </P ><P > Creating shared libraries is generally analogous to linking executables: first the source files are compiled into object files, then the object files are linked together. The object files need to be created as <I CLASS="FIRSTTERM" >position-independent code</I > (<ACRONYM CLASS="ACRONYM" >PIC</ACRONYM >), which conceptually means that they can be placed at an arbitrary location in memory when they are loaded by the executable. (Object files intended for executables are usually not compiled that way.) The command to link a shared library contains special flags to distinguish it from linking an executable (at least in theory — on some systems the practice is much uglier). </P ><P > In the following examples we assume that your source code is in a file <TT CLASS="FILENAME" >foo.c</TT > and we will create a shared library <TT CLASS="FILENAME" >foo.so</TT >. The intermediate object file will be called <TT CLASS="FILENAME" >foo.o</TT > unless otherwise noted. A shared library can contain more than one object file, but we only use one here. </P ><P ></P ><DIV CLASS="VARIABLELIST" ><DL ><DT ><SPAN CLASS="SYSTEMITEM" >FreeBSD</SPAN ></DT ><DD ><P > The compiler flag to create <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is <TT CLASS="OPTION" >-fPIC</TT >. To create shared libraries the compiler flag is <TT CLASS="OPTION" >-shared</TT >. </P><PRE CLASS="PROGRAMLISTING" >gcc -fPIC -c foo.c gcc -shared -o foo.so foo.o</PRE ><P> This is applicable as of version 3.0 of <SPAN CLASS="SYSTEMITEM" >FreeBSD</SPAN >. </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >HP-UX</SPAN ></DT ><DD ><P > The compiler flag of the system compiler to create <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is <TT CLASS="OPTION" >+z</TT >. When using <SPAN CLASS="APPLICATION" >GCC</SPAN > it's <TT CLASS="OPTION" >-fPIC</TT >. The linker flag for shared libraries is <TT CLASS="OPTION" >-b</TT >. So: </P><PRE CLASS="PROGRAMLISTING" >cc +z -c foo.c</PRE ><P> or: </P><PRE CLASS="PROGRAMLISTING" >gcc -fPIC -c foo.c</PRE ><P> and then: </P><PRE CLASS="PROGRAMLISTING" >ld -b -o foo.sl foo.o</PRE ><P> <SPAN CLASS="SYSTEMITEM" >HP-UX</SPAN > uses the extension <TT CLASS="FILENAME" >.sl</TT > for shared libraries, unlike most other systems. </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >IRIX</SPAN ></DT ><DD ><P > <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is the default, no special compiler options are necessary. The linker option to produce shared libraries is <TT CLASS="OPTION" >-shared</TT >. </P><PRE CLASS="PROGRAMLISTING" >cc -c foo.c ld -shared -o foo.so foo.o</PRE ><P> </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >Linux</SPAN ></DT ><DD ><P > The compiler flag to create <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is <TT CLASS="OPTION" >-fPIC</TT >. The compiler flag to create a shared library is <TT CLASS="OPTION" >-shared</TT >. A complete example looks like this: </P><PRE CLASS="PROGRAMLISTING" >cc -fPIC -c foo.c cc -shared -o foo.so foo.o</PRE ><P> </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >Mac OS X</SPAN ></DT ><DD ><P > Here is an example. It assumes the developer tools are installed. </P><PRE CLASS="PROGRAMLISTING" >cc -c foo.c cc -bundle -flat_namespace -undefined suppress -o foo.so foo.o</PRE ><P> </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >NetBSD</SPAN ></DT ><DD ><P > The compiler flag to create <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is <TT CLASS="OPTION" >-fPIC</TT >. For <ACRONYM CLASS="ACRONYM" >ELF</ACRONYM > systems, the compiler with the flag <TT CLASS="OPTION" >-shared</TT > is used to link shared libraries. On the older non-ELF systems, <TT CLASS="LITERAL" >ld -Bshareable</TT > is used. </P><PRE CLASS="PROGRAMLISTING" >gcc -fPIC -c foo.c gcc -shared -o foo.so foo.o</PRE ><P> </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >OpenBSD</SPAN ></DT ><DD ><P > The compiler flag to create <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is <TT CLASS="OPTION" >-fPIC</TT >. <TT CLASS="LITERAL" >ld -Bshareable</TT > is used to link shared libraries. </P><PRE CLASS="PROGRAMLISTING" >gcc -fPIC -c foo.c ld -Bshareable -o foo.so foo.o</PRE ><P> </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >Solaris</SPAN ></DT ><DD ><P > The compiler flag to create <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is <TT CLASS="OPTION" >-KPIC</TT > with the Sun compiler and <TT CLASS="OPTION" >-fPIC</TT > with <SPAN CLASS="APPLICATION" >GCC</SPAN >. To link shared libraries, the compiler option is <TT CLASS="OPTION" >-G</TT > with either compiler or alternatively <TT CLASS="OPTION" >-shared</TT > with <SPAN CLASS="APPLICATION" >GCC</SPAN >. </P><PRE CLASS="PROGRAMLISTING" >cc -KPIC -c foo.c cc -G -o foo.so foo.o</PRE ><P> or </P><PRE CLASS="PROGRAMLISTING" >gcc -fPIC -c foo.c gcc -G -o foo.so foo.o</PRE ><P> </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >Tru64 UNIX</SPAN ></DT ><DD ><P > <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is the default, so the compilation command is the usual one. <TT CLASS="COMMAND" >ld</TT > with special options is used to do the linking. </P><PRE CLASS="PROGRAMLISTING" >cc -c foo.c ld -shared -expect_unresolved '*' -o foo.so foo.o</PRE ><P> The same procedure is used with GCC instead of the system compiler; no special options are required. </P ></DD ><DT ><SPAN CLASS="SYSTEMITEM" >UnixWare</SPAN ></DT ><DD ><P > The compiler flag to create <ACRONYM CLASS="ACRONYM" >PIC</ACRONYM > is <TT CLASS="OPTION" >-K PIC</TT > with the SCO compiler and <TT CLASS="OPTION" >-fpic</TT > with <SPAN CLASS="PRODUCTNAME" >GCC</SPAN >. To link shared libraries, the compiler option is <TT CLASS="OPTION" >-G</TT > with the SCO compiler and <TT CLASS="OPTION" >-shared</TT > with <SPAN CLASS="PRODUCTNAME" >GCC</SPAN >. </P><PRE CLASS="PROGRAMLISTING" >cc -K PIC -c foo.c cc -G -o foo.so foo.o</PRE ><P> or </P><PRE CLASS="PROGRAMLISTING" >gcc -fpic -c foo.c gcc -shared -o foo.so foo.o</PRE ><P> </P ></DD ></DL ></DIV ><DIV CLASS="TIP" ><BLOCKQUOTE CLASS="TIP" ><P ><B >Tip: </B > If this is too complicated for you, you should consider using <A HREF="http://www.gnu.org/software/libtool/" TARGET="_top" > <SPAN CLASS="PRODUCTNAME" >GNU Libtool</SPAN ></A >, which hides the platform differences behind a uniform interface. </P ></BLOCKQUOTE ></DIV ><P > The resulting shared library file can then be loaded into <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN >. When specifying the file name to the <TT CLASS="COMMAND" >CREATE FUNCTION</TT > command, one must give it the name of the shared library file, not the intermediate object file. Note that the system's standard shared-library extension (usually <TT CLASS="LITERAL" >.so</TT > or <TT CLASS="LITERAL" >.sl</TT >) can be omitted from the <TT CLASS="COMMAND" >CREATE FUNCTION</TT > command, and normally should be omitted for best portability. </P ><P > Refer back to <A HREF="xfunc-c.html#XFUNC-C-DYNLOAD" >Section 35.9.1</A > about where the server expects to find the shared library files. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="AEN53419" >35.9.7. Composite-type Arguments</A ></H2 ><P > Composite types do not have a fixed layout like C structures. Instances of a composite type can contain null fields. In addition, composite types that are part of an inheritance hierarchy can have different fields than other members of the same inheritance hierarchy. Therefore, <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > provides a function interface for accessing fields of composite types from C. </P ><P > Suppose we want to write a function to answer the query: </P><PRE CLASS="PROGRAMLISTING" >SELECT name, c_overpaid(emp, 1500) AS overpaid FROM emp WHERE name = 'Bill' OR name = 'Sam';</PRE ><P> Using call conventions version 0, we can define <CODE CLASS="FUNCTION" >c_overpaid</CODE > as: </P><PRE CLASS="PROGRAMLISTING" >#include "postgres.h" #include "executor/executor.h" /* for GetAttributeByName() */ #ifdef PG_MODULE_MAGIC PG_MODULE_MAGIC; #endif bool c_overpaid(HeapTupleHeader t, /* the current row of emp */ int32 limit) { bool isnull; int32 salary; salary = DatumGetInt32(GetAttributeByName(t, "salary", &isnull)); if (isnull) return false; return salary > limit; }</PRE ><P> In version-1 coding, the above would look like this: </P><PRE CLASS="PROGRAMLISTING" >#include "postgres.h" #include "executor/executor.h" /* for GetAttributeByName() */ #ifdef PG_MODULE_MAGIC PG_MODULE_MAGIC; #endif PG_FUNCTION_INFO_V1(c_overpaid); Datum c_overpaid(PG_FUNCTION_ARGS) { HeapTupleHeader t = PG_GETARG_HEAPTUPLEHEADER(0); int32 limit = PG_GETARG_INT32(1); bool isnull; Datum salary; salary = GetAttributeByName(t, "salary", &isnull); if (isnull) PG_RETURN_BOOL(false); /* Alternatively, we might prefer to do PG_RETURN_NULL() for null salary. */ PG_RETURN_BOOL(DatumGetInt32(salary) > limit); }</PRE ><P> </P ><P > <CODE CLASS="FUNCTION" >GetAttributeByName</CODE > is the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > system function that returns attributes out of the specified row. It has three arguments: the argument of type <TT CLASS="TYPE" >HeapTupleHeader</TT > passed into the function, the name of the desired attribute, and a return parameter that tells whether the attribute is null. <CODE CLASS="FUNCTION" >GetAttributeByName</CODE > returns a <TT CLASS="TYPE" >Datum</TT > value that you can convert to the proper data type by using the appropriate <CODE CLASS="FUNCTION" >DatumGet<TT CLASS="REPLACEABLE" ><I >XXX</I ></TT >()</CODE > macro. Note that the return value is meaningless if the null flag is set; always check the null flag before trying to do anything with the result. </P ><P > There is also <CODE CLASS="FUNCTION" >GetAttributeByNum</CODE >, which selects the target attribute by column number instead of name. </P ><P > The following command declares the function <CODE CLASS="FUNCTION" >c_overpaid</CODE > in SQL: </P><PRE CLASS="PROGRAMLISTING" >CREATE FUNCTION c_overpaid(emp, integer) RETURNS boolean AS '<TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT >/funcs', 'c_overpaid' LANGUAGE C STRICT;</PRE ><P> Notice we have used <TT CLASS="LITERAL" >STRICT</TT > so that we did not have to check whether the input arguments were NULL. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="AEN53443" >35.9.8. Returning Rows (Composite Types)</A ></H2 ><P > To return a row or composite-type value from a C-language function, you can use a special API that provides macros and functions to hide most of the complexity of building composite data types. To use this API, the source file must include: </P><PRE CLASS="PROGRAMLISTING" >#include "funcapi.h"</PRE ><P> </P ><P > There are two ways you can build a composite data value (henceforth a <SPAN CLASS="QUOTE" >"tuple"</SPAN >): you can build it from an array of Datum values, or from an array of C strings that can be passed to the input conversion functions of the tuple's column data types. In either case, you first need to obtain or construct a <TT CLASS="STRUCTNAME" >TupleDesc</TT > descriptor for the tuple structure. When working with Datums, you pass the <TT CLASS="STRUCTNAME" >TupleDesc</TT > to <CODE CLASS="FUNCTION" >BlessTupleDesc</CODE >, and then call <CODE CLASS="FUNCTION" >heap_form_tuple</CODE > for each row. When working with C strings, you pass the <TT CLASS="STRUCTNAME" >TupleDesc</TT > to <CODE CLASS="FUNCTION" >TupleDescGetAttInMetadata</CODE >, and then call <CODE CLASS="FUNCTION" >BuildTupleFromCStrings</CODE > for each row. In the case of a function returning a set of tuples, the setup steps can all be done once during the first call of the function. </P ><P > Several helper functions are available for setting up the needed <TT CLASS="STRUCTNAME" >TupleDesc</TT >. The recommended way to do this in most functions returning composite values is to call: </P><PRE CLASS="PROGRAMLISTING" >TypeFuncClass get_call_result_type(FunctionCallInfo fcinfo, Oid *resultTypeId, TupleDesc *resultTupleDesc)</PRE ><P> passing the same <TT CLASS="LITERAL" >fcinfo</TT > struct passed to the calling function itself. (This of course requires that you use the version-1 calling conventions.) <TT CLASS="VARNAME" >resultTypeId</TT > can be specified as <TT CLASS="LITERAL" >NULL</TT > or as the address of a local variable to receive the function's result type OID. <TT CLASS="VARNAME" >resultTupleDesc</TT > should be the address of a local <TT CLASS="STRUCTNAME" >TupleDesc</TT > variable. Check that the result is <TT CLASS="LITERAL" >TYPEFUNC_COMPOSITE</TT >; if so, <TT CLASS="VARNAME" >resultTupleDesc</TT > has been filled with the needed <TT CLASS="STRUCTNAME" >TupleDesc</TT >. (If it is not, you can report an error along the lines of <SPAN CLASS="QUOTE" >"function returning record called in context that cannot accept type record"</SPAN >.) </P ><DIV CLASS="TIP" ><BLOCKQUOTE CLASS="TIP" ><P ><B >Tip: </B > <CODE CLASS="FUNCTION" >get_call_result_type</CODE > can resolve the actual type of a polymorphic function result; so it is useful in functions that return scalar polymorphic results, not only functions that return composites. The <TT CLASS="VARNAME" >resultTypeId</TT > output is primarily useful for functions returning polymorphic scalars. </P ></BLOCKQUOTE ></DIV ><DIV CLASS="NOTE" ><BLOCKQUOTE CLASS="NOTE" ><P ><B >Note: </B > <CODE CLASS="FUNCTION" >get_call_result_type</CODE > has a sibling <CODE CLASS="FUNCTION" >get_expr_result_type</CODE >, which can be used to resolve the expected output type for a function call represented by an expression tree. This can be used when trying to determine the result type from outside the function itself. There is also <CODE CLASS="FUNCTION" >get_func_result_type</CODE >, which can be used when only the function's OID is available. However these functions are not able to deal with functions declared to return <TT CLASS="STRUCTNAME" >record</TT >, and <CODE CLASS="FUNCTION" >get_func_result_type</CODE > cannot resolve polymorphic types, so you should preferentially use <CODE CLASS="FUNCTION" >get_call_result_type</CODE >. </P ></BLOCKQUOTE ></DIV ><P > Older, now-deprecated functions for obtaining <TT CLASS="STRUCTNAME" >TupleDesc</TT >s are: </P><PRE CLASS="PROGRAMLISTING" >TupleDesc RelationNameGetTupleDesc(const char *relname)</PRE ><P> to get a <TT CLASS="STRUCTNAME" >TupleDesc</TT > for the row type of a named relation, and: </P><PRE CLASS="PROGRAMLISTING" >TupleDesc TypeGetTupleDesc(Oid typeoid, List *colaliases)</PRE ><P> to get a <TT CLASS="STRUCTNAME" >TupleDesc</TT > based on a type OID. This can be used to get a <TT CLASS="STRUCTNAME" >TupleDesc</TT > for a base or composite type. It will not work for a function that returns <TT CLASS="STRUCTNAME" >record</TT >, however, and it cannot resolve polymorphic types. </P ><P > Once you have a <TT CLASS="STRUCTNAME" >TupleDesc</TT >, call: </P><PRE CLASS="PROGRAMLISTING" >TupleDesc BlessTupleDesc(TupleDesc tupdesc)</PRE ><P> if you plan to work with Datums, or: </P><PRE CLASS="PROGRAMLISTING" >AttInMetadata *TupleDescGetAttInMetadata(TupleDesc tupdesc)</PRE ><P> if you plan to work with C strings. If you are writing a function returning set, you can save the results of these functions in the <TT CLASS="STRUCTNAME" >FuncCallContext</TT > structure — use the <TT CLASS="STRUCTFIELD" >tuple_desc</TT > or <TT CLASS="STRUCTFIELD" >attinmeta</TT > field respectively. </P ><P > When working with Datums, use: </P><PRE CLASS="PROGRAMLISTING" >HeapTuple heap_form_tuple(TupleDesc tupdesc, Datum *values, bool *isnull)</PRE ><P> to build a <TT CLASS="STRUCTNAME" >HeapTuple</TT > given user data in Datum form. </P ><P > When working with C strings, use: </P><PRE CLASS="PROGRAMLISTING" >HeapTuple BuildTupleFromCStrings(AttInMetadata *attinmeta, char **values)</PRE ><P> to build a <TT CLASS="STRUCTNAME" >HeapTuple</TT > given user data in C string form. <TT CLASS="LITERAL" >values</TT > is an array of C strings, one for each attribute of the return row. Each C string should be in the form expected by the input function of the attribute data type. In order to return a null value for one of the attributes, the corresponding pointer in the <TT CLASS="PARAMETER" >values</TT > array should be set to <TT CLASS="SYMBOL" >NULL</TT >. This function will need to be called again for each row you return. </P ><P > Once you have built a tuple to return from your function, it must be converted into a <TT CLASS="TYPE" >Datum</TT >. Use: </P><PRE CLASS="PROGRAMLISTING" >HeapTupleGetDatum(HeapTuple tuple)</PRE ><P> to convert a <TT CLASS="STRUCTNAME" >HeapTuple</TT > into a valid Datum. This <TT CLASS="TYPE" >Datum</TT > can be returned directly if you intend to return just a single row, or it can be used as the current return value in a set-returning function. </P ><P > An example appears in the next section. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="XFUNC-C-RETURN-SET" >35.9.9. Returning Sets</A ></H2 ><P > There is also a special API that provides support for returning sets (multiple rows) from a C-language function. A set-returning function must follow the version-1 calling conventions. Also, source files must include <TT CLASS="FILENAME" >funcapi.h</TT >, as above. </P ><P > A set-returning function (<ACRONYM CLASS="ACRONYM" >SRF</ACRONYM >) is called once for each item it returns. The <ACRONYM CLASS="ACRONYM" >SRF</ACRONYM > must therefore save enough state to remember what it was doing and return the next item on each call. The structure <TT CLASS="STRUCTNAME" >FuncCallContext</TT > is provided to help control this process. Within a function, <TT CLASS="LITERAL" >fcinfo->flinfo->fn_extra</TT > is used to hold a pointer to <TT CLASS="STRUCTNAME" >FuncCallContext</TT > across calls. </P><PRE CLASS="PROGRAMLISTING" >typedef struct { /* * Number of times we've been called before * * call_cntr is initialized to 0 for you by SRF_FIRSTCALL_INIT(), and * incremented for you every time SRF_RETURN_NEXT() is called. */ uint32 call_cntr; /* * OPTIONAL maximum number of calls * * max_calls is here for convenience only and setting it is optional. * If not set, you must provide alternative means to know when the * function is done. */ uint32 max_calls; /* * OPTIONAL pointer to result slot * * This is obsolete and only present for backward compatibility, viz, * user-defined SRFs that use the deprecated TupleDescGetSlot(). */ TupleTableSlot *slot; /* * OPTIONAL pointer to miscellaneous user-provided context information * * user_fctx is for use as a pointer to your own data to retain * arbitrary context information between calls of your function. */ void *user_fctx; /* * OPTIONAL pointer to struct containing attribute type input metadata * * attinmeta is for use when returning tuples (i.e., composite data types) * and is not used when returning base data types. It is only needed * if you intend to use BuildTupleFromCStrings() to create the return * tuple. */ AttInMetadata *attinmeta; /* * memory context used for structures that must live for multiple calls * * multi_call_memory_ctx is set by SRF_FIRSTCALL_INIT() for you, and used * by SRF_RETURN_DONE() for cleanup. It is the most appropriate memory * context for any memory that is to be reused across multiple calls * of the SRF. */ MemoryContext multi_call_memory_ctx; /* * OPTIONAL pointer to struct containing tuple description * * tuple_desc is for use when returning tuples (i.e., composite data types) * and is only needed if you are going to build the tuples with * heap_form_tuple() rather than with BuildTupleFromCStrings(). Note that * the TupleDesc pointer stored here should usually have been run through * BlessTupleDesc() first. */ TupleDesc tuple_desc; } FuncCallContext;</PRE ><P> </P ><P > An <ACRONYM CLASS="ACRONYM" >SRF</ACRONYM > uses several functions and macros that automatically manipulate the <TT CLASS="STRUCTNAME" >FuncCallContext</TT > structure (and expect to find it via <TT CLASS="LITERAL" >fn_extra</TT >). Use: </P><PRE CLASS="PROGRAMLISTING" >SRF_IS_FIRSTCALL()</PRE ><P> to determine if your function is being called for the first or a subsequent time. On the first call (only) use: </P><PRE CLASS="PROGRAMLISTING" >SRF_FIRSTCALL_INIT()</PRE ><P> to initialize the <TT CLASS="STRUCTNAME" >FuncCallContext</TT >. On every function call, including the first, use: </P><PRE CLASS="PROGRAMLISTING" >SRF_PERCALL_SETUP()</PRE ><P> to properly set up for using the <TT CLASS="STRUCTNAME" >FuncCallContext</TT > and clearing any previously returned data left over from the previous pass. </P ><P > If your function has data to return, use: </P><PRE CLASS="PROGRAMLISTING" >SRF_RETURN_NEXT(funcctx, result)</PRE ><P> to return it to the caller. (<TT CLASS="LITERAL" >result</TT > must be of type <TT CLASS="TYPE" >Datum</TT >, either a single value or a tuple prepared as described above.) Finally, when your function is finished returning data, use: </P><PRE CLASS="PROGRAMLISTING" >SRF_RETURN_DONE(funcctx)</PRE ><P> to clean up and end the <ACRONYM CLASS="ACRONYM" >SRF</ACRONYM >. </P ><P > The memory context that is current when the <ACRONYM CLASS="ACRONYM" >SRF</ACRONYM > is called is a transient context that will be cleared between calls. This means that you do not need to call <CODE CLASS="FUNCTION" >pfree</CODE > on everything you allocated using <CODE CLASS="FUNCTION" >palloc</CODE >; it will go away anyway. However, if you want to allocate any data structures to live across calls, you need to put them somewhere else. The memory context referenced by <TT CLASS="STRUCTFIELD" >multi_call_memory_ctx</TT > is a suitable location for any data that needs to survive until the <ACRONYM CLASS="ACRONYM" >SRF</ACRONYM > is finished running. In most cases, this means that you should switch into <TT CLASS="STRUCTFIELD" >multi_call_memory_ctx</TT > while doing the first-call setup. </P ><P > A complete pseudo-code example looks like the following: </P><PRE CLASS="PROGRAMLISTING" >Datum my_set_returning_function(PG_FUNCTION_ARGS) { FuncCallContext *funcctx; Datum result; <TT CLASS="REPLACEABLE" ><I >further declarations as needed</I ></TT > if (SRF_IS_FIRSTCALL()) { MemoryContext oldcontext; funcctx = SRF_FIRSTCALL_INIT(); oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx); /* One-time setup code appears here: */ <TT CLASS="REPLACEABLE" ><I >user code</I ></TT > <TT CLASS="REPLACEABLE" ><I >if returning composite</I ></TT > <TT CLASS="REPLACEABLE" ><I >build TupleDesc, and perhaps AttInMetadata</I ></TT > <TT CLASS="REPLACEABLE" ><I >endif returning composite</I ></TT > <TT CLASS="REPLACEABLE" ><I >user code</I ></TT > MemoryContextSwitchTo(oldcontext); } /* Each-time setup code appears here: */ <TT CLASS="REPLACEABLE" ><I >user code</I ></TT > funcctx = SRF_PERCALL_SETUP(); <TT CLASS="REPLACEABLE" ><I >user code</I ></TT > /* this is just one way we might test whether we are done: */ if (funcctx->call_cntr < funcctx->max_calls) { /* Here we want to return another item: */ <TT CLASS="REPLACEABLE" ><I >user code</I ></TT > <TT CLASS="REPLACEABLE" ><I >obtain result Datum</I ></TT > SRF_RETURN_NEXT(funcctx, result); } else { /* Here we are done returning items and just need to clean up: */ <TT CLASS="REPLACEABLE" ><I >user code</I ></TT > SRF_RETURN_DONE(funcctx); } }</PRE ><P> </P ><P > A complete example of a simple <ACRONYM CLASS="ACRONYM" >SRF</ACRONYM > returning a composite type looks like: </P><PRE CLASS="PROGRAMLISTING" >PG_FUNCTION_INFO_V1(retcomposite); Datum retcomposite(PG_FUNCTION_ARGS) { FuncCallContext *funcctx; int call_cntr; int max_calls; TupleDesc tupdesc; AttInMetadata *attinmeta; /* stuff done only on the first call of the function */ if (SRF_IS_FIRSTCALL()) { MemoryContext oldcontext; /* create a function context for cross-call persistence */ funcctx = SRF_FIRSTCALL_INIT(); /* switch to memory context appropriate for multiple function calls */ oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx); /* total number of tuples to be returned */ funcctx->max_calls = PG_GETARG_UINT32(0); /* Build a tuple descriptor for our result type */ if (get_call_result_type(fcinfo, NULL, &tupdesc) != TYPEFUNC_COMPOSITE) ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), errmsg("function returning record called in context " "that cannot accept type record"))); /* * generate attribute metadata needed later to produce tuples from raw * C strings */ attinmeta = TupleDescGetAttInMetadata(tupdesc); funcctx->attinmeta = attinmeta; MemoryContextSwitchTo(oldcontext); } /* stuff done on every call of the function */ funcctx = SRF_PERCALL_SETUP(); call_cntr = funcctx->call_cntr; max_calls = funcctx->max_calls; attinmeta = funcctx->attinmeta; if (call_cntr < max_calls) /* do when there is more left to send */ { char **values; HeapTuple tuple; Datum result; /* * Prepare a values array for building the returned tuple. * This should be an array of C strings which will * be processed later by the type input functions. */ values = (char **) palloc(3 * sizeof(char *)); values[0] = (char *) palloc(16 * sizeof(char)); values[1] = (char *) palloc(16 * sizeof(char)); values[2] = (char *) palloc(16 * sizeof(char)); snprintf(values[0], 16, "%d", 1 * PG_GETARG_INT32(1)); snprintf(values[1], 16, "%d", 2 * PG_GETARG_INT32(1)); snprintf(values[2], 16, "%d", 3 * PG_GETARG_INT32(1)); /* build a tuple */ tuple = BuildTupleFromCStrings(attinmeta, values); /* make the tuple into a datum */ result = HeapTupleGetDatum(tuple); /* clean up (this is not really necessary) */ pfree(values[0]); pfree(values[1]); pfree(values[2]); pfree(values); SRF_RETURN_NEXT(funcctx, result); } else /* do when there is no more left */ { SRF_RETURN_DONE(funcctx); } }</PRE ><P> One way to declare this function in SQL is: </P><PRE CLASS="PROGRAMLISTING" >CREATE TYPE __retcomposite AS (f1 integer, f2 integer, f3 integer); CREATE OR REPLACE FUNCTION retcomposite(integer, integer) RETURNS SETOF __retcomposite AS '<TT CLASS="REPLACEABLE" ><I >filename</I ></TT >', 'retcomposite' LANGUAGE C IMMUTABLE STRICT;</PRE ><P> A different way is to use OUT parameters: </P><PRE CLASS="PROGRAMLISTING" >CREATE OR REPLACE FUNCTION retcomposite(IN integer, IN integer, OUT f1 integer, OUT f2 integer, OUT f3 integer) RETURNS SETOF record AS '<TT CLASS="REPLACEABLE" ><I >filename</I ></TT >', 'retcomposite' LANGUAGE C IMMUTABLE STRICT;</PRE ><P> Notice that in this method the output type of the function is formally an anonymous <TT CLASS="STRUCTNAME" >record</TT > type. </P ><P > The directory <A HREF="tablefunc.html" >contrib/tablefunc</A > module in the source distribution contains more examples of set-returning functions. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="AEN53566" >35.9.10. Polymorphic Arguments and Return Types</A ></H2 ><P > C-language functions can be declared to accept and return the polymorphic types <TT CLASS="TYPE" >anyelement</TT >, <TT CLASS="TYPE" >anyarray</TT >, <TT CLASS="TYPE" >anynonarray</TT >, <TT CLASS="TYPE" >anyenum</TT >, and <TT CLASS="TYPE" >anyrange</TT >. See <A HREF="extend-type-system.html#EXTEND-TYPES-POLYMORPHIC" >Section 35.2.5</A > for a more detailed explanation of polymorphic functions. When function arguments or return types are defined as polymorphic types, the function author cannot know in advance what data type it will be called with, or need to return. There are two routines provided in <TT CLASS="FILENAME" >fmgr.h</TT > to allow a version-1 C function to discover the actual data types of its arguments and the type it is expected to return. The routines are called <TT CLASS="LITERAL" >get_fn_expr_rettype(FmgrInfo *flinfo)</TT > and <TT CLASS="LITERAL" >get_fn_expr_argtype(FmgrInfo *flinfo, int argnum)</TT >. They return the result or argument type OID, or <TT CLASS="SYMBOL" >InvalidOid</TT > if the information is not available. The structure <TT CLASS="LITERAL" >flinfo</TT > is normally accessed as <TT CLASS="LITERAL" >fcinfo->flinfo</TT >. The parameter <TT CLASS="LITERAL" >argnum</TT > is zero based. <CODE CLASS="FUNCTION" >get_call_result_type</CODE > can also be used as an alternative to <CODE CLASS="FUNCTION" >get_fn_expr_rettype</CODE >. </P ><P > For example, suppose we want to write a function to accept a single element of any type, and return a one-dimensional array of that type: </P><PRE CLASS="PROGRAMLISTING" >PG_FUNCTION_INFO_V1(make_array); Datum make_array(PG_FUNCTION_ARGS) { ArrayType *result; Oid element_type = get_fn_expr_argtype(fcinfo->flinfo, 0); Datum element; bool isnull; int16 typlen; bool typbyval; char typalign; int ndims; int dims[MAXDIM]; int lbs[MAXDIM]; if (!OidIsValid(element_type)) elog(ERROR, "could not determine data type of input"); /* get the provided element, being careful in case it's NULL */ isnull = PG_ARGISNULL(0); if (isnull) element = (Datum) 0; else element = PG_GETARG_DATUM(0); /* we have one dimension */ ndims = 1; /* and one element */ dims[0] = 1; /* and lower bound is 1 */ lbs[0] = 1; /* get required info about the element type */ get_typlenbyvalalign(element_type, &typlen, &typbyval, &typalign); /* now build the array */ result = construct_md_array(&element, &isnull, ndims, dims, lbs, element_type, typlen, typbyval, typalign); PG_RETURN_ARRAYTYPE_P(result); }</PRE ><P> </P ><P > The following command declares the function <CODE CLASS="FUNCTION" >make_array</CODE > in SQL: </P><PRE CLASS="PROGRAMLISTING" >CREATE FUNCTION make_array(anyelement) RETURNS anyarray AS '<TT CLASS="REPLACEABLE" ><I >DIRECTORY</I ></TT >/funcs', 'make_array' LANGUAGE C IMMUTABLE;</PRE ><P> </P ><P > There is a variant of polymorphism that is only available to C-language functions: they can be declared to take parameters of type <TT CLASS="LITERAL" >"any"</TT >. (Note that this type name must be double-quoted, since it's also a SQL reserved word.) This works like <TT CLASS="TYPE" >anyelement</TT > except that it does not constrain different <TT CLASS="LITERAL" >"any"</TT > arguments to be the same type, nor do they help determine the function's result type. A C-language function can also declare its final parameter to be <TT CLASS="LITERAL" >VARIADIC "any"</TT >. This will match one or more actual arguments of any type (not necessarily the same type). These arguments will <SPAN CLASS="emphasis" ><I CLASS="EMPHASIS" >not</I ></SPAN > be gathered into an array as happens with normal variadic functions; they will just be passed to the function separately. The <CODE CLASS="FUNCTION" >PG_NARGS()</CODE > macro and the methods described above must be used to determine the number of actual arguments and their types when using this feature. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="XFUNC-TRANSFORM-FUNCTIONS" >35.9.11. Transform Functions</A ></H2 ><P > Some function calls can be simplified during planning based on properties specific to the function. For example, <TT CLASS="LITERAL" >int4mul(n, 1)</TT > could be simplified to just <TT CLASS="LITERAL" >n</TT >. To define such function-specific optimizations, write a <I CLASS="FIRSTTERM" >transform function</I > and place its OID in the <TT CLASS="STRUCTFIELD" >protransform</TT > field of the primary function's <TT CLASS="STRUCTNAME" >pg_proc</TT > entry. The transform function must have the SQL signature <TT CLASS="LITERAL" >protransform(internal) RETURNS internal</TT >. The argument, actually <TT CLASS="TYPE" >FuncExpr *</TT >, is a dummy node representing a call to the primary function. If the transform function's study of the expression tree proves that a simplified expression tree can substitute for all possible concrete calls represented thereby, build and return that simplified expression. Otherwise, return a <TT CLASS="LITERAL" >NULL</TT > pointer (<SPAN CLASS="emphasis" ><I CLASS="EMPHASIS" >not</I ></SPAN > a SQL null). </P ><P > We make no guarantee that <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > will never call the primary function in cases that the transform function could simplify. Ensure rigorous equivalence between the simplified expression and an actual call to the primary function. </P ><P > Currently, this facility is not exposed to users at the SQL level because of security concerns, so it is only practical to use for optimizing built-in functions. </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="AEN53612" >35.9.12. Shared Memory and LWLocks</A ></H2 ><P > Add-ins can reserve LWLocks and an allocation of shared memory on server startup. The add-in's shared library must be preloaded by specifying it in <A HREF="runtime-config-resource.html#GUC-SHARED-PRELOAD-LIBRARIES" >shared_preload_libraries</A >. Shared memory is reserved by calling: </P><PRE CLASS="PROGRAMLISTING" >void RequestAddinShmemSpace(int size)</PRE ><P> from your <CODE CLASS="FUNCTION" >_PG_init</CODE > function. </P ><P > LWLocks are reserved by calling: </P><PRE CLASS="PROGRAMLISTING" >void RequestAddinLWLocks(int n)</PRE ><P> from <CODE CLASS="FUNCTION" >_PG_init</CODE >. </P ><P > To avoid possible race-conditions, each backend should use the LWLock <CODE CLASS="FUNCTION" >AddinShmemInitLock</CODE > when connecting to and initializing its allocation of shared memory, as shown here: </P><PRE CLASS="PROGRAMLISTING" >static mystruct *ptr = NULL; if (!ptr) { bool found; LWLockAcquire(AddinShmemInitLock, LW_EXCLUSIVE); ptr = ShmemInitStruct("my struct name", size, &found); if (!found) { initialize contents of shmem area; acquire any requested LWLocks using: ptr->mylockid = LWLockAssign(); } LWLockRelease(AddinShmemInitLock); }</PRE ><P> </P ></DIV ><DIV CLASS="SECT2" ><H2 CLASS="SECT2" ><A NAME="EXTEND-CPP" >35.9.13. Using C++ for Extensibility</A ></H2 ><P > Although the <SPAN CLASS="PRODUCTNAME" >PostgreSQL</SPAN > backend is written in C, it is possible to write extensions in C++ if these guidelines are followed: <P ></P ></P><UL ><LI ><P > All functions accessed by the backend must present a C interface to the backend; these C functions can then call C++ functions. For example, <TT CLASS="LITERAL" >extern C</TT > linkage is required for backend-accessed functions. This is also necessary for any functions that are passed as pointers between the backend and C++ code. </P ></LI ><LI ><P > Free memory using the appropriate deallocation method. For example, most backend memory is allocated using <CODE CLASS="FUNCTION" >palloc()</CODE >, so use <CODE CLASS="FUNCTION" >pfree()</CODE > to free it. Using C++ <CODE CLASS="FUNCTION" >delete</CODE > in such cases will fail. </P ></LI ><LI ><P > Prevent exceptions from propagating into the C code (use a catch-all block at the top level of all <TT CLASS="LITERAL" >extern C</TT > functions). This is necessary even if the C++ code does not explicitly throw any exceptions, because events like out-of-memory can still throw exceptions. Any exceptions must be caught and appropriate errors passed back to the C interface. If possible, compile C++ with <TT CLASS="OPTION" >-fno-exceptions</TT > to eliminate exceptions entirely; in such cases, you must check for failures in your C++ code, e.g. check for NULL returned by <CODE CLASS="FUNCTION" >new()</CODE >. </P ></LI ><LI ><P > If calling backend functions from C++ code, be sure that the C++ call stack contains only plain old data structures (<ACRONYM CLASS="ACRONYM" >POD</ACRONYM >). This is necessary because backend errors generate a distant <CODE CLASS="FUNCTION" >longjmp()</CODE > that does not properly unroll a C++ call stack with non-POD objects. </P ></LI ></UL ><P> </P ><P > In summary, it is best to place C++ code behind a wall of <TT CLASS="LITERAL" >extern C</TT > functions that interface to the backend, and avoid exception, memory, and call stack leakage. </P ></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="xfunc-internal.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="xaggr.html" ACCESSKEY="N" >Next</A ></TD ></TR ><TR ><TD WIDTH="33%" ALIGN="left" VALIGN="top" >Internal Functions</TD ><TD WIDTH="34%" ALIGN="center" VALIGN="top" ><A HREF="extend.html" ACCESSKEY="U" >Up</A ></TD ><TD WIDTH="33%" ALIGN="right" VALIGN="top" >User-defined Aggregates</TD ></TR ></TABLE ></DIV ></BODY ></HTML >