Provided by: libpqtypes-dev_1.5.1-5_amd64 bug


       pqt-specs - A manual for libpqtypes data type specifier strings.


       The type system allows addressing backend data types by their fully qualified schema name.
       Similar to the backend type system, libpqtypes has an input and output function  for  each
       type:  put  and  get  respectively.   All  builtin  types  are  supported  by  libpqtypes.
       Additional types, such as composites and user-defined types, can be registered by an  API-
       user on a per connection basis.

       Putting  and  getting  types  are  addressed by their backend names as printf-style format
       specifiers.  For instance: %int4, %timestamptz or %text.  They can also be addressed using
       their   fully   qualified   schema   names:   %pg_catalog.int4,   %pg_catalog.timestamptz,
       %pg_catalog.text or %myschema.mytype to avoid ambiguity.

   Specifier Strings
       Specifier Mark
       A specifier mark indicates where a specifier begins. A mark is not part of the type name.

          %  Marks the beginning of a type name.  When getting results, this
             also indicates that a field will be referenced by number.

          #  Marks the beginning of a type name.  When getting results, this
             also indicates that a field will be referenced by name.

          @  Marks the beginning of a prepared specifier name, see
             PQspecPrepare().  When used, it must be the first and only
             specifier in the format string: "@prepared_spec", "@myfunc",
             etc...  NOTE: the ´@´ must be the first character, no spaces.

       Type Specifier
       Type specifiers are comprised of an optional schema name and type name.   Type  specifiers
       have a set of rules:

          -) Format: [schema].type - optional schema name, a "." separator
             between schema and type and the type name.

          -) First character must be a-z or an underscore

          -) Double quotes are required for characters not in [a-zA-Z0-9_]
             NOTE: In libpqtypes, this includes "double precision"

          -) Schema "." separator, specifier marks or flags are not included
             in double quotes

          -) Non-quoted type names are casefolded, quoted names are not.

          -) Examples:
             "%\"my oDd~ !tYpe naMe#\""
             "%myschema.\"my oDd~ !tYpe naMe#\""
             "%abc.int4 %pg_catalog.int4" <= fully qualified type names

          -) Last in First out: To find a type referenced in a specifier
             string, the search begins with the last type registered.
             User registered type handlers are searched first, followed
             by builtin types.

          -) pqt schema: There is a builtin schema named pqt.  By default,
             it contains two types: ´str´ and ´null´.  Anything can be
             put into this schema, which has nothing to do with the server.
             This is good for aliases or type sub-classes that are

       Specifier Flag
       Flags  are  used to alter the behavior of a type specifier.  They are always placed at the
       end of the specifier name.  If the name is double quoted,  the  flag  is  just  after  the
       closing quote.

          *  This is called the pointer flag.  It is only supported on a
             handful of builtin types during a PQputf(3), but user registered
             types can provide support for them.  Supported types are:
             VARCHAR, BPCHAR, TEXT, BYTEA, NAME and the pqt.str.

             Putting data: this flag tells libpqtypes to store a direct
             pointer to the data being put, rather than making a copy of

             Getting data: no built-in types make use of the pointer flag.
             User-defined type handlers can make the pointer flag behave
             anyway they see fit.  The ´get´ type handler is supplied a
             PGtypeArgs which contains an ´is_ptr´ member.

          [] This is called the array flag.  It indicates that an array is
             being referenced rather than a simple type.  This flag is always
             used with a PGarray.


       With  the  exception  of  the  "numeric" type, all numeric types behave identically: int2,
       int4, int8, float4 and float8.

              PG type    C type
              PGint2     short
              PGint4     int
              PGint8     long long (platform dependent)
              PGfloat4   float
              PGfloat8   double

       Putting numeric values: If the value supplied is too large for the PG  type,  it  will  be
       silently truncated.
              PQputf(param, "%int2 %int4 %int8 %float4 %float8",
                   SHRT_MAX, INT_MAX, LLONG_MAX, 1234.56, 123456.789);

       Getting  numeric  values:  Like  scanf,  the  correctly sized data type must be used.  For
       instance: you cannot use a 4-byte int for %int2 - you must use a short.
              // Read an int2 from field 0, int4 from field 1, int8 from
              // field 2, float4 from field 3 and a float8 from field 4
              PGint2 i2;
              PGint4 i4;
              PGint8 i8;
              PGfloat4 f4;
              PGfloat8 f8;
              PQgetf(result, tup_num, "%int2 %int4 %int8 %float4 %float8",
                   0, &i2, 1, &i4, 2, &i8, 3, &f4, 4, &f8);

       The numeric data type is always exposed in text format.  There is no C data structure.  It
       is always sent in binary format.
              PQputf(param, "%numeric", "1728718718271827121233.121212121212");

       Even  if  binary results are used when getting a numeric value, libpqtypes will internally
       convert the numeric to text. This has the advantage of allowing you to use binary  results
       and  still  have  access  to numeric fields.  If you want to work with a numeric in binary
       form, use PQgetvalue() on a binary result set.
              PGnumeric numstr;
              PQgetf(result, tup_num, "%numeric", field_num, &numstr);

       The first argument is the field number of the numeric. The second argument is a pointer to
       a PGnumeric to receive the numeric string value (which will always be NUL terminated).


       Arrays  are  put  using  the  PGarray  structure.  Elements are put using PQputf(3) into a
       PGparam structure contained withn a PGarray.  The PGarray contains array dimension members
       for  specifing  the  number  of  dimension, the dimensions of each dimension and the lower
       bounds of each dimension.

       Arrays are only handled using binary format.  This means that any type used  as  an  array
       element  must  be  put  and  gotten  in  binary  format.   If a user-defined type does not
       implement a send and recv function in the backend, it can not be used as an array element.

       For a discussion of composite arrays, `man pqt-composites(3)´.
              typedef struct
                   /* The number of array dimensions.  Specifing zero for this
                    * value on puts has special meaning.  When zero, this value
                    * is set to one, dims[0] is set to the number of items in
                    * the ´param´ member and lbound[0] is set to one.
                   int ndims;

                   /* An array of lower bounds for each dimension. */
                   int lbound[MAXDIM];

                   /* An array of dimensions for each dimension. */
                   int dims[MAXDIM];

                   /* When putting array elements, this PGparam is used.  Each
                    * item put is one array element.  Because the PQputf(3)
                    * interface allows putting more than one item at a time, you
                    * can put multiple array elements.
                   PGparam *param;

                   /* When getting an array, this is the PGresult object that
                    * contains the array elements.  Each element is one tuple,
                    * regardless of the array dimensions.  If the array has 100
                    * elements across 3 dimensions, PQntuples(arr.res) will return
                    * 100.  The only valid field, for non-composite arrays, is
                    * field zero.
                   PGresult *res;
              } PGarray;

       When all elements have been put, the PGarray structure must be put using  the  "[]"  array
       specifer flag into a PGparam structure.  PQputf(3) is used to build the array elements and
       to put the resulting PGarray.

       Putting an array value:
              PGint4 i;
              PGarray arr;
              PGparam *param;

              /* One dimensional arrays do not require setting dimension info. For
               * convience, you can zero the structure or set ndims to zero.
              arr.ndims = 0;

              /* create the param object that will contain the elements */
              arr.param = PQparamCreate(conn);

              /* Use PQputf(3) to put the array elements */
              for(i=0; i < 1000; i++)
                PQputf(arr.param, "%int4", i);

              /* The PGarray must be put into a PGparam struture.  So far, only
               * the array elements have been put.  ´param´ can continue to be
               * used to pack more parameters.  The array is now a single parameter
               * within ´param´.
              param = PQparamCreate(conn);
              PQputf(param, "%int[]", &arr);

              /* no longer needed */

       To get an array, PQgetf(3) is used in conjunction with the PGarray structure.  The  result
       object  contained  with  the  PGarray is populated with the array elements.  The dimension
       info is assigned as well.  Each array element is its own tuple that only contains a single
       field for non composite arrays.

       Getting an array value:
              int i;
              PGint4 val;
              int ntups;
              PGarray arr;
              PGresult *result = ...;

              PQgetf(result, 0, "%int[]", 0, &arr);

              /* not needed anymore*/

              /* each tuple is an array element */
              ntups = PQntuples(arr.res);
              for(i=0; i < ntups; i++)
                /* Always field 0 */
                PQgetf(arr.res, i, "%int4", 0, &val);
                printf("[%03d] %d\n", i, val);


       The  result object is not organized based on the dimension info.  Indexes are always zero-
       based.  If the dimension info is meaningful to your application, index translation must be
       done using the ndims, dims and lbound members of the PGarray structure.

       You  cannot  reference an array element by field name.  The only logical name for an array
       element would be the string version of its index ... "0", "1", etc..  The index value of a
       non-composite  array  is  its  tuple  number, the field number is always zero.  This means
       "#int" is not legal for non-composite arrays.  You must use "%int" and reference field 0.


       The "char" data type uses the PGchar.  The value is limited to 8-bits.

       Putting a "char" value:
              PGchar c = ´a´;
              PQputf(param, "%char %char", 213, c);

       Getting a "char" value:
              PGchar c;
              PQgetf(result, tup_num, "%char", field_num, &c);


       BPCHAR, VARCHAR, NAME and TEXT are handled identically.  libpqtypes does no range checking
       on  NAME,  BPCHAR or VARCHAR, it lets the server perform these checks.  There are two ways
       to put strings: allow libpqtypes to make an internal copy of the string (default behavior)
       or as a direct pointer: (both require that the C string is NUL-terminated)
              /* Put a string so libpqtypes makes a copy. In this case,
               * 4 copies would be made of the same string.
              PGtext str = "foobar";
              PQputf(param, "%bpchar %varchar %name %text", str, str, str, str);

              /* Put a string so libpqtypes doesn´t make a copy,
               * keeps a direct pointer.  More efficient than above,
               * especially if these are large strings.
              PQputf(param, "%bpchar* %varchar* %name* %text*", str, str, str, str);

       WARNING: Be careful about variable scope when using the "*" specifier flag:
              /* when ´func´ returns, the str pointer becomes invalid!
               * The below should be using "%text" ... w/o the * flag.
              int func(PGparam *param)
                   PGchar str[16];
                   strcpy(str, "foobar");
                   return PQputf(param, "%text*", str); // BAD IDEA!

       To  PQgetf(3)  a  string, you supply a pointer to a PGtext.  Unlike putting string values,
       getting them doesn´t make use of the "*" specifier flag (silently ignored).
              /* Get a string value */
              PGvarchar str;
              PQgetf(result, tup_num, "%varchar", field_num, &str);

              /* identical to */
              str = PQgetvalue(result, tup_num, field_num);

       The reason the ´*´ specifier flag is silently ignored, rather than raising a syntax error,
       is it is common to define macros for specifer strings; that can be used for puts and gets:
              /* user_id, username, password */
              #define TYPESPEC_USERINFO "%int4 %text* %text*"

              PGint4 uid = 0;
              PGtext user = "foo", pass = "bar";
              PQputf(param, TYPESPEC_USERINFO, uid, user, pass);
              PQgetf(param, tup_num, TYPESPEC_USERINFO, 0, &uid, 1, &user, 2, &pass);

       The above allowance is more useful than a syntax error.


       There  are  two  ways  to  put  a bytea: copy or direct pointer (just like variable-length
       character types).  In either case, you supply a pointer to a PGbytea.
              typedef struct
                int len;    /* number of bytes */
                char *data; /* pointer to the bytea data */
              } PGbytea;

              /* Put a bytea letting libpqtypes make a copy; */
              PGbytea bytea = {4, {0, 1, 2, 3}};
              PQputf(param, "%bytea", &bytea);

              /* Put a bytea not letting libpqtypes make a copy, stores a
               * direct pointer to
              PQputf(param, "%bytea*", &bytea);

       To get a bytea, you provide a pointer to a PGbytea.  Unlike putting bytea values, there is
       only one way to get them.
              /* Get a bytea value (exposed as binary, no
               * escaping/unescaping needed)
              PGbytea bytea;
              PQgetf(result, tup_num, "%bytea", field_num, &bytea);

       NOTE:  For text results, PQgetlength will not match the length returned by PQgetf(3).  The
       values PQgetf(3) assigns to the user  provided  PGbytea*  represent  the  unescaped  bytea


       PGdate  is used by DATE, TIMESTAMP and TIMESTAMPTZ data types. To put a date, you must set
       the isbc, year, mon and mday members.  All other members are ignored.

              typedef struct
                   /* When non-zero, the date is in the BC ERA. */
                   int isbc;

                    * The BC or AD year, which is NOT adjusted by 1900 like
                    * the POSIX struct tm.  Years are always positive values,
                    * even BC years.  To distinguish between BC and AD years,
                    * use the isbc flag: (year 0 not used)
                    *   Ex. -1210 is represented as: isbc=1, year=1209
                   int year;

                   /* The number of months since January, in the range 0 to 11. */
                   int mon;

                   /* The day of the month, in the range 1 to 31. */
                   int mday;

                   /* The Julian day in the Gregorian calendar. */
                   int jday;

                   /* The number of days since January 1, in the range 0 to 365. */
                   int yday;

                   /* The number of days since Sunday, in the range 0 to 6. */
                   int wday;
              } PGdate;

       Putting a date value:
              // ´1401-01-19 BC´
              PGdate date;
              date.isbc = 1;
              date.year = 1401;
              date.mon  = 0;
              date.mday = 19;
              PQputf(param, "%date", &date);

       Getting a date value:
              PQgetf(result, tup_num, "%date", field_num, &date);


       PGtime is used by TIME, TIMETZ, TIMESTAMP and TIMESTAMPTZ data types. To put a  time,  you
       must set the hour, min, sec and usec members.  All other members are ignored.
              typedef struct
                   /* The number of hours past midnight, in the range 0 to 23. */
                   int hour;

                   /* The number of minutes after the hour, in the
                    * range 0 to 59.
                   int min;

                   /* The number of seconds after the minute, in the
                    * range 0 to 59.
                   int sec;

                   /* The number of microseconds after the second, in the
                    * range of 0 to 999999.
                   int usec;

                    * When non-zero, this is a TIME WITH TIME ZONE.  Otherwise,
                    * it is a TIME WITHOUT TIME ZONE.
                   int withtz;

                   /* A value of 1 indicates daylight savings time.  A value of 0
                    * indicates standard time.  A value of -1 means unknown or
                    * could not determine.
                   int isdst;

                   /* Offset from UTC in seconds. This value is not always
                    * available. It is set to 0 if it cannot be determined.
                   int gmtoff;

                   /* Timezone abbreviation: such as EST, GMT, PDT, etc.
                    * This value is not always available.  It is set to an empty
                    * string if it cannot be determined.
                   char tzabbr[16];
              } PGtime;

       Putting a time value:
              // ´10:41:06.002897´
              PGdate time;
              time.hour   = 10;
              time.min    = 41;
              time.sec    = 6;
              time.usec   = 2897;
              PQputf(param, "%time", &time);

       Getting a time value:
              PQgetf(result, tup_num, "%time", field_num, &time);


       The  TIMETZ  data  type uses the PGtime structure, for a description of this structure see
       the TIME section.  To put a timetz, you must set the  hour,  min,  sec,  usec  and  gmtoff
       members.  All other members are ignored.

       Putting a timetz value:
              // ´10:41:06.002897-05´
              PGdate timetz;
              timetz.hour   = 10;
              timetz.min    = 41;
              timetz.sec    = 6;
              timetz.usec   = 2897;
              timetz.gmtoff = -18000;
              PQputf(param, "%timetz", &timetz);

       Getting a timetz value:
              PQgetf(result, tup_num, "%timetz", field_num, &timetz);


       To  put  a  timestamp,  the isbc, year, mon, mday, hour, min, sec and usec members must be
       set.  No other members are used.
              typedef struct
                   /* The number seconds before or after midnight UTC of
                    * January 1, 1970, not counting leap seconds.
                   PGint8 epoch;

                   /* The date part of the timestamp. */
                   PGdate date;

                   /* The time part of the timestamp. */
                   PGtime time;
              } PGtimestamp;

       Putting a timestamp value:
              // ´2000-01-19 10:41:06´
              PGtimestamp ts;
       = 0;
       = 2000;
        = 0;
       = 19;
              ts.time.hour   = 10;
              ts.time.min    = 41;
              ts.time.sec    = 6;
              ts.time.usec   = 0;
              PQputf(param, "%timestamp", &ts);

       Getting a timestamp value:
              PQgetf(result, tup_num, "%timestamp", field_num, &ts);

       The timestamp type has no concept of timezone, so  the  value  returned  by  PQgetf(3)  is
       exactly what the server sent; no timezone adjustments are attempted.  The gmtoff is always
       set to zero, tzabbr will be an empty string and withtz will be zero.


       To put a timestamptz, the isbc, year, mon, mday, hour, min, sec, usec and  gmtoff  members
       must be set.  No other members are used.

       Putting a timestamptz value:
              // ´2000-01-19 10:41:06-05´
              PGtimestamp ts;
       = 0;
       = 2000;
        = 0;
       = 19;
              ts.time.hour   = 10;
              ts.time.min    = 41;
              ts.time.sec    = 6;
              ts.time.usec   = 0;
              ts.time.gmtoff = -18000;
              PQputf(param, "%timestamptz", &ts);

       Getting a timestamptz value:
              PQgetf(result, tup_num, "%timestamptz", field_num, &ts);

       When  using  PQgetf(3)  in  binary mode, the timestamptz value is converted into the local
       machine´s timezone.  If the local machine's timezone can not be determined, the value will
       be in GMT (gmtoff is set to zero and tzabbr is set to GMT).

       When  using PQgetf(3) in text mode, the timestamptz value is returned as a datetime string
       in the server´s timezone.  No adjustments are made to this value.  If the server is  using
       a  DateStyle  that encodes the gmtoff "00:00:00-05", then gmtoff will be set to this value
       and tzabbr will be "GMT+/-hhmmss" (00:00:00-05 => GMT-0500).  In this case, isdst  is  set
       to  -1  ...  meaning  unknown.  If the server´s DateStyle encodes a timezone abbreviation,
       like PST, then tzabbr is set to this value.  The gmtoff and  isdst  members  are  properly

         DateStyle includes a timezone abbrev - "SQL, MDY"
         01/25/2007 00:00:00 EST => tzabbr=EST, gmtoff=-18000, isdst=0
         01/25/2007 01:00:00 EDT => tzabbr=EDT, gmtoff=-14400, isdst=1


       To  put an interval, all relevant members of a PGinterval should be assigned and those not
       used should be set to zero.
              typedef struct
                   /* the number of years */
                   int years;

                   /* the number of months */
                   int mons;

                   /* the number of days */
                   int days;

                   /* the number of hours */
                   int hours;

                   /* the number of mins */
                   int mins;

                   /* the number of seconds */
                   int secs;

                   /* the number of microseconds */
                   int usecs;
              } PGinterval;

       Putting an interval value:
              // "20 years 8 months 9 hours 10 mins 15 secs 123456 usecs"
              PGinterval interval;
              interval.years = 20;
              interval.mons  = 8;
              interval.days  = 0; // not used, set to 0
              interval.hours = 9;
              interval.mins  = 10;
              interval.secs  = 15;
              interval.usecs = 123456;
              PQputf(param, "%interval", &interval);

       Getting an interval value:
              PQgetf(result, tup_num, "%interval", field_num, &interval);

       NOTE: When using text results with a non-ISO DateStyle, microseconds are truncated to a  2
       digit value.  For example: "4 mins 2.11 secs" but microseconds is really 111456.


       The PGpoint structure is used to put and get a point.
              typedef struct
                   double x; // point x value
                   double y; // point y value
              } PGpoint;

       Putting a point value:
              PGpoint pt = {12.345, 6.789};
              PQputf(param, "%point", &pt);

       Getting a point value:
              PGpoint pt;
              PQgetf(result, tup_num, "%point", field_num, &pt);


       The PGlseg structure is used to put and get a line segnment.
              typedef struct
                PGpoint pts[2];
              } PGlseg;

       Putting a lseg value:
              PGlseg lseg = {{{12.345, 6.789}, {99.8, 88.9}}};
              PQputf(param, "%lseg", &lseg);

       Getting a lseg value:
              PGlseg lseg;
              PQgetf(result, tup_num, "%lseg", field_num, &lseg);


       The PGbox structure is used to put and get a box.
              typedef struct
                PGpoint high;
                PGpoint low;
              } PGbox;

       Putting a box value:
              PGbox box = {{12.345, 6.789}, {22.234, 1.9998}};
              PQputf(param, "%box", &box);

       Getting a box value:
              PGbox box;
              PQgetf(result, tup_num, "%box", field_num, &box);


       The PGcircle structure is used to put and get a circle.
              typedef struct
                PGpoint center;
                double radius;
              } PGcircle;

       Putting a circle value:
              PGcircle circle = {{12.345, 6.789}, 2.34567};
              PQputf(param, "%circle", &circle);

       Getting a circle value:
              PGcircle circle;
              PQgetf(result, tup_num, "%circle", field_num, &circle);


       The PGpath structure is used to put and get a path.  If the closed member is non-zero, the
       path is closed, otherwise it is open.
              typedef struct
                int npts;
                int closed;
                PGpoint *pts;
              } PGpath;

       Putting a path value:
              // Put a closed path that contains 2 points
              PGpoint pts[] = {{12.345, 6.789}, {19.773, 7.882}};
              PGpath path = {2, 1, pts};
              PQputf(param, "%path", &path);

       Getting a path value:
              PGpath path;
              if(PQgetf(result, tup_num, "%path", field_num, &path))
                   // path.pts must be copied out if needed after clearing results
                   copy_points(path.npts, path.pts, ...);

                   // path.pts is now invalid!


       The PGpolygon structure is used to put and get a polygon.
              typedef struct
                int npts;
                PGpoint *pts;
              } PGpolygon;

       Putting a polygon value:
              // Put a polygon that contains 2 points
              PGpoint pts[] = {{12.345, 6.789}, {19.773, 7.882}};
              PGpolygon polygon = {2, 1, pts};
              PQputf(param, "%polygon", &polygon);

       Getting a polygon value:
              PGpolygon polygon;
              if(PQgetf(result, tup_num, "%polygon", field_num, &polygon))
                   // polygon.pts must be copied out if needed after clearing results
                   copy_points(polygon.npts, polygon.pts, ...);

                   // polygon.pts is now invalid


       When putting an inet or cidr, all members must be set excluding the sa_len.
              typedef struct
                   /* The address mask, 32 for a single IP. */
                   int mask;

                   /* When non-zero, the PGinet structure represents a cidr
                    * otherwise an inet.
                   int is_cidr;

                   /* the length in bytes of the sa_buf member. */
                   int sa_len;

                   /* the socket address buffer, contains the data.  This can
                    * be casted to a sockaddr, sockaddr_in, sockaddr_in6 or a
                    * sockaddr_storage structure. This buffer is 128 bytes so
                    * that it is large enough for a sockaddr_storage structure.
                   char sa_buf[128];
              } PGinet;

       Putting an inet or cidr:
              socklen_t len;
              PGinet inet;

              cli_fd = accept(srv_fd, (struct sockaddr *)inet.sa_buf, &len);
              if(cli_fd != -1)
                   inet.is_cidr = 0;
                   inet.mask = 32;
                   PQputf(param, "%inet", &inet);

       Getting an inet or cidr:
              PGinet inet;
              unsigned short port;

              /* gets an inet from field 2 and an int2 from field 6 */
              if(PQgetf(result, tup_num, "%inet %int2", 2, &inet, 6, &port))
                   char ip[80];
                   struct sockaddr *sa = (struct sockaddr *)inet.sa_buf;

                   // converting a PGinet to an IPv4 or IPv6 address string
                   getnameinfo(sa, inet.sa_len, ip, sizeof(ip),
                        NULL, 0, NI_NUMERICHOST);

                   // The inet data type does not store a port.
                   if(sa->sa_family == AF_INET)
                        ((struct sockaddr_in *)sa)->sin_port = htons(port);
                        ((struct sockaddr_in6 *)sa)->sin6_port = htons(port);

                   printf("Connecting to %s:%d\n", ip, port);
                   connect(sock_fd, sa, inet.sa_len);


       The PGmacaddr structure is used to put and get a macaddr.
              typedef struct
                   int a;
                   int b;
                   int c;
                   int d;
                   int e;
                   int f;
              } PGmacaddr;

       Putting a macaddr value:
              PGmacaddr macaddr = {0, 1, 2, 3, 4, 5};
              PQputf(param, "%macaddr", &macaddr);

       Getting a macaddr value:
              PGmacaddr macaddr;
              PQgetf(result, tup_num, "%macaddr", field_num, &macaddr);


       The money type is put/get as a PGmoney (64-bit integer).  It can be  converted  to  dollar
       and  cents  format  by  dividing by 100: double money = (double)money64 / 100.0;.  Pre 8.3
       servers are limited to 32-bit money values.

       Putting a money value:
              PGmoney money = 600000000054LL; // 6 billion dollars and 54 cents
              PQputf(param, "%money", money);

       Getting a money value:
              PQgetf(result, tup_num, "%money", field_num, &money);


       The bool type is put/get as a PGbool.  To put true or false, use 1 or 0.

       Putting a bool value:
              PGbool b = 1; // put true
              PQputf(param, "%bool", b);

       Getting a bool value:
              PGbool b;
              PQgetf(result, tup_num, "%bool", field_num, &b);


       The uuid type is put/get as a sequence of 16 bytes.  To put a uuid as  text,  use  "%str".
       NOTE: this type is not available on pre 8.3 servers.

       Putting a uuid value:
              PGuuid uuid = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
              PQputf(param, "%uuid", uuid);

       Getting a uuid value:
              PGuuid uuid;
              PQgetf(result, tup_num, "%uuid", field_num, &uuid);

       WARNING: The data provided on a put call is expected to be at least 16 bytes.


       Putting an oid value:
              Oid oid = 2318;
              PQputf(param, "%oid", oid);

       Getting an oid value:
              Oid oid;
              PQgetf(result, tup_num, "%oid", field_num, &oid);




       A  contribution  of eSilo, LLC. for the PostgreSQL Database Management System.  Written by
       Andrew Chernow and Merlin Moncure.


       Report bugs to <>.


       Copyright (c) 2011 eSilo, LLC. All rights reserved.
       This is free software; see the source for copying conditions.  There is NO  warranty;  not


       PQgetf(3), PQputf(3).