Re: [PyDB2-discuss] Re: Db2 module issues.

[Yuval T. <yu...@il...>]

??? wrote:

>>3.  In DB2_connection_object_cursor (line 666 :-), there's a commented 
>>out call to Py_INCREF(self) right after the assignment of cursor->conn = 
>>self.  If we want to make sure that the connection object stays alive 
>>while the cursor is alive, we must call incref ( and decref upon 
>>destruction ).
>>    
>>
>
>I didn't find the clean way to avoid cyclic reference situation.
>In case of db.close(), users expect cursors from that ``db'' also close,
>I think.  How about your idea?
>
Ok - I saw that we're holding a reference to the connection object 
inside the python cursor object.  So there's really no need to hold it 
in the C module as well.  Also, I saw that the only function that uses 
this connection object from within the C cursor is 'callproc'.  So what 
I did is removed the C connection reference from the C cursor (thus no 
need to play with ref. counting at all), and in 
DB2_cursor_object_callproc I added/changed the following lines:

    DB2ConnectionObject *conn = NULL;

    /* start parsing */
    if ( !PyArg_ParseTuple(args, "Os|O", conn, &proc_name, &p_args) ) {
        return NULL;
    }

    if ( ! conn )
    {
        PyErr_SetString(DB2_Error, "Connection object not found.");
        return NULL;
    }

And in the callproc method of the python cursor I changed the following 
lines:

    def callproc(self, procname, *args):
        t_args = args[:]
   
        r = self._cs.callproc(self._conn, procname.upper(), t_args)
        self._refresh_info()

This way, we're holding the ref. of the connection object only once - in 
the python code, and we send this reference to the C module only when we 
need to use it (only in callproc at the moment).

What do you say ?

-- 
Yuval Turgeman
Content Technology, Aduva LTD.