Preparing to write C code
Babelfish can be accessed using pretty much every programming language known to man. This logically includes C / C++. In this section, you will learn how to write simple C programs to access Babelfish.
All examples shown in this section have been written for and on Linux using the FreeTDS library, which is readily available for all widespread Linux distributions. To get started, it’s important to install the FreeTDS driver.
On Fedora Linux this works as follows (as of Fedora Core 34):
[root@fedora ~]# dnf install freetds*
Last metadata expiration check: 0:02:52 ago on Fr 24 Sep 2021 11:11:43 CEST.
Dependencies resolved.
====================================================================================================================================================
Package Architecture Version Repository Size
====================================================================================================================================================
Installing:
freetds x86_64 1.1.20-4.fc34 fedora 373 k
freetds-devel x86_64 1.1.20-4.fc34 fedora 39 k
freetds-doc noarch 1.1.20-4.fc34 fedora 1.0 M
Installing dependencies:
freetds-libs x86_64 1.1.20-4.fc34 fedora 423 k
If you are coding in C, make sure that the -devel packages (= header files) are also installed. These header files will be needed to compile the code on your system.
Connecting to the database
The first thing we want to achieve is to connect to the database and check the connection. The following listing contains a small C program to make that happen:
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/param.h>
#include <sybfront.h>
#include <sybdb.h>
#include <syberror.h>
#define UID "postgres" // Information censored
#define PWD "verysecret!"
#define PROGNAME "demo_prog"
#define DBSERVER "sample-host.example.com"
#define DBNAME "postgres"
/* handler from messages from the server */
static int
msg_handler(DBPROCESS* dbproc, DBINT msgno, int msgstate, int severity,
char *msgtext, char *srvname, char *procname, int line)
{
/* regular errors are handled by the error handler */
if (severity < 11)
fprintf(stderr, "Server message (severity %d): %s\n", severity, msgtext);
return 0;
}
/* error handler */
static int err_handler(DBPROCESS* dbproc, int severity, int dberr, int oserr, char *dberrstr, char *oserrstr)
{
fprintf(stderr, "Server error %d: %s\n", dberr, dberrstr);
if (oserr != 0)
fprintf(stderr, "Caused by system error %d: %s\n", oserr, oserrstr);
return INT_CANCEL;
}
int main(void)
{
LOGINREC *login;
DBPROCESS *dbconn;
char hostname[MAXHOSTNAMELEN];
int max_len = MAXHOSTNAMELEN;
DBCHAR accession[10];
DBCHAR examdesc[10];
DBCHAR examcode[255];
char portstr[20];
int rc;
if (dbinit() == FAIL)
{
fprintf(stderr, "Could not init db.\n");
return 1;
}
/* Allocate a login params structure */
if ((login = dblogin()) == FAIL)
{
fprintf(stderr, "Could not initialize dblogin() structure.\n");
return 2;
}
/* Initialize the login params in the structure */
DBSETLUSER(login, UID);
DBSETLPWD(login, PWD);
// DBSETLAPP(login, PROGNAME);
if (gethostname(hostname, max_len) == 0)
{
DBSETLHOST(login, hostname);
fprintf(stderr, "setting login hostname: %s\n", hostname);
}
/* the port can only be set via environment variable */
rc = snprintf(portstr, 20, "TDSPORT=%d", 1433);
if (rc < 0 || rc >= 20)
{
fprintf(stderr, "error composing string for environment variable TDSPORT\n");
return 0;
}
if (putenv(portstr) != 0)
{
fprintf(stderr, "error setting TDSPORT environment variable\n");
return 0;
}
/* install error handler */
dberrhandle(err_handler);
dbmsghandle(msg_handler);
/* Now connect to the DB Server */
if ((dbconn = dbopen(login, DBSERVER)) == NULL)
{
fprintf(stderr, "Could not connect to DB Server: %s\n", DBSERVER);
return 3;
}
printf("success\n");
return 0;
}
In our example, we have compiled the code using GCC:
gcc main_01.c -lsybdb -I/usr/include/ -o main_01
Two things are of crucial importance to compile the code: The -I flag tells the compiler where to look for the header files. -lsydbd ensures that the FreeTDS library is properly linked. The binary created is called main_01.
The first thing the code does is to include some header files. What we see here are sybfront.h, sybdb.h and syberror.h. In the next step, we define message and error handlers. Without these functions, the code can be executed normally, but the user experience won’t be what people usually expect from a proper application. That’s why it makes sense to have such functions around. In our case, all they do is to send messages to stderr. However, it is up to you how you want to handle the log.
Then we can already get into the structures and continue the creation of our connection. It is usually not strictly necessary to set the desired port as an environment variable, but we have found it useful to do that in many cases. An example how to achieve that has been included in this sample.
Finally the connection is opened (dbopen) and we check for errors. Overall more code is needed than in the case of a simple libpq program, but it is not too complicated either.
What can go wrong?
FreeTDS does not want to work out of the box without a configuration file. In case things go wrong, the error message might look like this:
[hs@fedora tds_test]$ ./main_01
setting login hostname: fedora
Server error 20018: General SQL Server error: Check messages from the SQL Server
Caused by system error -1: (null)
Server error 20002: Adaptive Server connection failed (sample-host.example.com)
Server error 20002: Adaptive Server connection failed (sample-host.example.com)
Could not connect to DB Server: sample-host.example.com)
In this case, the FreeTDS config is wrong or simply doesn’t exist in the desired form. The error message is pretty obscure, therefore it is important to understand where to look for the error.
Configuring FreeTDS
Please see https://www.freetds.org/userguide/freetdsconf.html