PQexecPrepared() behavior

From: Tatsuo Ishii <ishii(at)postgresql(dot)org>
To: pgsql-hackers(at)postgresql(dot)org
Subject: PQexecPrepared() behavior
Date: 2009-08-30 12:01:55
Message-ID: 20090830.210155.70221428.t-ishii@sraoss.co.jp
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

While looking into a pgpool-II problem reported by a user, I found
weird behavior of PQexecPrepared().

Strange thing is, it seems PQexecPrepared() sends B(bind), Describe,
Execute and Sync at once without checking the result of Bind
message. Is this leagal from a point of view of the frontend/backend
protocol? PostgreSQL version is 8.4.0.

Here is the trace.

To backend> Msg Q
To backend> "begin;"
To backend> Msg complete, length 12
From backend> C
From backend (#4)> 10
From backend> "BEGIN"
From backend> Z
From backend (#4)> 5
From backend> Z
From backend (#4)> 5
From backend> T
Preparing the statement...
To backend> Msg P
To backend> ""
To backend> "INSERT INTO test_table (a) VALUES ($1)"
To backend (2#)> 1
To backend (4#)> 0
To backend> Msg complete, length 51
To backend> Msg S
To backend> Msg complete, length 5
From backend> 1
From backend (#4)> 4
From backend> Z
From backend (#4)> 5
From backend> Z
From backend (#4)> 5
From backend> T
Executing the statement...
To backend> Msg B <-- Bind
To backend> ""
To backend> ""
To backend (2#)> 0
To backend (2#)> 1
To backend (4#)> 5
To backend> 123.5
To backend (2#)> 1
To backend (2#)> 0
To backend> Msg complete, length 24
To backend> Msg D <-- Describe
To backend> P
To backend> ""
To backend> Msg complete, length 7
To backend> Msg E <-- Execute
To backend> ""
To backend (4#)> 0
To backend> Msg complete, length 10
To backend> Msg S <-- Sync
To backend> Msg complete, length 5
From backend> E
From backend (#4)> 88
From backend> S
From backend> "ERROR"
From backend> C
From backend> "22P02"
From backend> M
From backend> "invalid input syntax for integer: "123.5""
From backend> F
From backend> "numutils.c"
From backend> L
From backend> "106"
From backend> R
From backend> "pg_atoi"
From backend>
From backend> Z
From backend (#4)> 5
From backend> Z
From backend (#4)> 5
From backend> E
Error executing the statement...

Error Message:ERROR: invalid input syntax for integer: "123.5"
To backend> Msg X
To backend> Msg complete, length 5

Here is the program provided by the user:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <unistd.h>

#include "libpq-fe.h"
#include "libpq/libpq-fs.h"

int main(int argc, char **argv) {
PGconn *conn;
PGresult *res;
Oid *param_types;
char **param_values;
int doTransaction = 1;

conn = PQconnectdb("user=t-ishii dbname=test port=5433");
if (PQstatus(conn) == CONNECTION_BAD) {
printf("Unable to connect to db\n");
PQfinish(conn);
return 1;
}

PQtrace(conn, stdout);

if(doTransaction == 1) {
PQexec(conn,"begin;");
}

printf("Preparing the statement...\n");
param_types = calloc(1,sizeof(Oid));
param_types[0] = 0;

param_values = calloc(1,sizeof(char *));
param_values[0] = "123.5";

res = PQprepare(conn,"","INSERT INTO test_table (a) VALUES ($1)",1,param_types);
switch(PQresultStatus(res)) {
case PGRES_COMMAND_OK:
case PGRES_TUPLES_OK:
break;
default:
printf("%s\nError Message:%s\n","Error preparing the statement...",PQresultErrorMessage(res));
PQclear(res);
free(param_types);
PQfinish(conn);
return 1;
break;
}

PQclear(res);

printf("Executing the statement...\n");
res = PQexecPrepared(conn,"",1,param_values,NULL,NULL,0);
switch(PQresultStatus(res)) {
case PGRES_COMMAND_OK:
case PGRES_TUPLES_OK:
break;
default:
printf("%s\nError Message:%s","Error executing the statement...\n",PQresultErrorMessage(res));
PQclear(res);
free(param_types);
PQfinish(conn);
return 1;
break;
}
if(doTransaction == 1) {
PQexec(conn,"commit;");
}

free(param_types);

PQfinish(conn);
return 0;
}
--
Tatsuo Ishii
SRA OSS, Inc. Japan

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Simon Riggs 2009-08-30 12:22:41 Re: LWLock Queue Jumping
Previous Message Grzegorz Jaskiewicz 2009-08-30 11:21:04 Re: clang's static checker report.