Hi all,

I have a simple procedure (basically it is a dummy procedure to test
the C-API interface)

CREATE
PROCEDURE get_day_info( p_sDate VARCHAR(8))
BEGIN
SELECT 0, 1, 1;
END

and can execute it inside the mysql terminal:

mysql> call get_day_info('xxx');
+---+---+---+
| 0 | 1 | 1 |
+---+---+---+
| 0 | 1 | 1 |
+---+---+---+
1 row in set (0.00 sec)

Query OK, 0 rows affected (0.01 sec)

I can also retrieve these results from the C-API with the sequence:

mysql_query( conn, "call get_day_info('xxx')");

result = mysql_use_result(conn);

while ( row = mysql_fetch_row(result) )
{
... do something with row[0] ... row[2]
}

mysql_free_result(result);

So far so good: All the 3 string values that are returned by the
procedure
are processed and in local buffers of my application.

The strange thing happens now, when I issue a:

mysql_next_result(conn);

It returns 0, which indicates that there are more result sets
available,
so I let a mysql_store_result() follow.
But mysql_store_result() returns the NULL pointer, while mysql_errno()
is
zero.

So why is 'mysql_next_result()' fooling me?

I am working with libmysqlclient version 5.0.45 on MS Windows 2000.
The database server is on a Unix machine and has version 5.0.32

Not sure, if this might cause some complications.
I'd rather expect, that it is related with the multithreading
architecture
of my application.

Here is a outline of my startup code:

run mysql_library_init() without errors

Checking with mysql_thread_safe() returns TRUE

open 3 sessions to the database server
each session with:

conn = mysql_init(NULL)
and
mysql_real_connect( ...., CLIENT_MULTI_RESULTS)

each session has its own Mutex to make sure, that always only one
thread
is using any of the mysql_something() API functions that work with
the
connection handle.

I'm happy about any comments or hints on this. At the moment I don't
see any harm, but frankly "Detecting the existence of a result" and
"Seeing it gone in the next moment" sounds a bit like a memory leak,
that strikes you back right in the moment, when you start to
deploy the application.

Thanks, Rolf.