Hello,

I've found a bug in mysql++, when a BLOB field is used which is not the
only field returned from a query. This happens only when the query is
run on a 64bit machine (AMD64 in my case). To demonstrate, use the following
table with the load_file and cgi_image examples:

CREATE TABLE fax ( id INTEGER AUTO_INCREMENT PRIMARY KEY,
fax BLOB );

execute load_file as usual.

Change cgi_image to return id and fax as a result of the select.

query << "SELECT " << MY_KEY << "," << MY_FIELD << " FROM " << MY_TABLE
<< " WHERE " << MY_KEY << " = " << argv[1];
ResUse res = query.use();
Row row = res.fetch_row();
long unsigned int *jj = res.fetch_lengths();
cout << "Content-length: " << jj[1] << endl << endl;
fwrite(row.raw_data(1), 1, jj[1], stdout);

jj[1] returns a proper length, but raw_data is not correct, and it looks
like a memory dump. I'm having difficulty debugging this, as I cannot
quite find where things go wrong.

Any ideas?

Grtz
Erwin

--
erwin (AT) leastwanted (DOT) nl
I pledge allegiance to the form of science fiction and fantasy and to
the promise for which it stands: one world, polycultural, genetically
engineered, with Internet access for all. -- Tonio Roque

--
MySQL++ Mailing List
For list archives: http://lists.mysql.com/plusplus
To unsubscribe: http://lists.mysql.com/plusplus?unsu...ie.nctu.edu.tw

On Sun, Mar 20, 2005 at 10:25:30PM +0100, Erwin wrote:
Two things to try:

try query.store() and see if anything changes.

Also, try the plain mysql C API to make sure that this is not a bug in
mysql. Here's some sample code to get you started. It compiles but I've
not run it here. It is based on some code posted earlier on the list.

g++ -o mysql_blob mysql_blob.cc -I/usr/include/mysql -lmysqlclient

mysql_blob.cc:

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

int main() {
int rc;
int i;
MYSQL *mysql;

mysql_server_init(0, NULL, NULL);

mysql = mysql_init(NULL);
if (mysql_real_connect(mysql, NULL, NULL, NULL, "", 0, NULL, 0) == NULL) {
printf("error\n");
}
if (mysql_errno(mysql)) {
printf("Error 1: %s\n", mysql_error(mysql));
mysql_close(mysql);
exit(-1);
}

if( mysql_query(mysql, "SELECT fax FROM fax WHERE id = 1") ) {
printf("Error: %s\n", mysql_error(mysql));
mysql_close(mysql);
exit(2);
}

MYSQL_RES *res = mysql_store_result(mysql);
if( res ) {
printf("Error: %s\n", mysql_error(mysql));
mysql_close(mysql);
exit(2);
}

MYSQL_ROW d = mysql_fetch_row(res);
if( d ) {
printf("Error: %s\n", mysql_error(mysql));
mysql_close(mysql);
exit(2);
}

unsigned long *len = mysql_fetch_lengths(res);

printf("Content-length: %lu\n", len[0]);
if( d[0] )
fwrite(d[0], 1, len[0], stdout);
else
printf("NULL!\n");

mysql_free_result(res);

mysql_close(mysql);
mysql_server_end();
}


--
MySQL++ Mailing List
For list archives: http://lists.mysql.com/plusplus
To unsubscribe: http://lists.mysql.com/plusplus?unsu...ie.nctu.edu.tw

Hello Chris,

Thanx, but I've solved the issue. There were some casts in connection.h
from unsigned long* to unsigned int* that went wrong. These casts were
already removed in 1.7.32 (I was using 1.7.28), so the issue has been
resolved.

Thank you for your time,
Erwin

--
erwin (AT) leastwanted (DOT) nl
We all enter the world in the same way: naked... screaming... soaked in
blood. But if you live your life right, that kind of thing doesn't have
to stop there.

--
MySQL++ Mailing List
For list archives: http://lists.mysql.com/plusplus
To unsubscribe: http://lists.mysql.com/plusplus?unsu...ie.nctu.edu.tw