[BUG] double conversion with ESQL/C on Linux
Hello!
We have experienced strange casting problem when storing doubles to
smallint on Informix/Linux. Here is the ESQL/C program that
demonstrates the error. It works OK on Sinix, but stores wrong
data on Linux.
Does anyone know where to obtain the patch for this bug?
Pozdrav,
Roman
- - -
/***********************************************************************
****
* $Id: testdb.ec,v 1.2 1999/08/18 11:57:14 romanm Exp $
*
* DESCRIPTION: Test file for strange Informix behavior regarding
conversion
* of C types to database types.
* We expect no change of value when inserting an integer value stored
* in a double variable to smallint DB-type.
*
* OUR ENVIRONMENT:
* - Linux guinness 2.2.5 #4 Tue Apr 13 16:51:36 MEST 1999 i686 unknown
* (SuSE 6.1),
* - Informix DS v7_30_uc7_2, ESQL 9.14 (11/11/93, 07:43:13)
* - GCC VERSION 2.7.2.3
* - compile and execute commands: esql testdb.ec; ./a.out
*
* OUR OUTPUT:
* ---->8----
* Testing SELECT type casting for prog01
* Expected value: 23, b_double=22.000000, b_int=23, delta:-1.000000
* Expected value: 24, b_double=23.000000, b_int=24, delta:-1.000000
* Expected value: 27, b_double=26.000000, b_int=27, delta:-1.000000
* Expected value: 28, b_double=27.000000, b_int=28, delta:-1.000000
* Expected value: 29, b_double=28.000000, b_int=29, delta:-1.000000
*
* Test is over.
* ---->8----
************************************************************************
**/
#include <stdio.h>
EXEC SQL BEGIN DECLARE SECTION;
int a_int, b_int;
double a_double, b_double;
EXEC SQL END DECLARE SECTION;
#define MAXVALUE 30
void prinerror(char *);
void printerror(char *action)
{
if (strncmp(SQLSTATE, "00", 2) != 0)
{
/* if (strncmp(SQLSTATE, "02", 2) != 0) */
printf("SQLSTATE after %s is %s\\n", action, SQLSTATE);
printf("SQLCODE is %d\\n", SQLCODE);
}
}
int main(void)
{
int value;
printf( "Testing SELECT type casting for prog01\\n");
EXEC SQL CONNECT TO 'prog01';
/* in case we have the table, we'll drop it */
EXEC SQL DROP TABLE tmp_foo_table;
EXEC SQL CREATE TABLE tmp_foo_table
(
x_int SMALLINT NOT NULL,
x_double SMALLINT NOT NULL
);
/* fill the table with values */
for (value=1; value<MAXVALUE; value++)
{
a_int = a_double = value;
EXEC SQL
INSERT INTO tmp_foo_table VALUES(:a_int, :a_double);
printerror("INSERT");
}
/* ERROR!! If we use dbaccess to view the contents of the
tmp_foo_table,
* there are already wrong values of x_double in it. */
/* read from table */
for (value=1; value<MAXVALUE; value++)
{
a_int = a_double = value;
EXEC SQL SELECT x_int, x_double
INTO :b_int, :b_double
FROM tmp_foo_table
WHERE x_int=:a_int;
if (b_double != b_int)
/* ERROR!! Sometimes this actually happens! */
printf("Expected value: %d, b_double=%f, b_int=%d, delta:%f\\n",
value, b_double, b_int, b_double-b_int);
printerror("SELECT");
}
/* EXEC SQL DROP TABLE tmp_foo_table; */
EXEC SQL DISCONNECT CURRENT;
printf("\\nTest is over.\\n");
return 0;
}
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.