[OpenSER-Devel] carrierroute

[Henning Westerholt]

On Wednesday 26 September 2007, Carsten Bock wrote:
> Hi,
>
> after understanding, how the carrier-route-module works (thanks
> Henning), i have a question regarding the source.
> I looked at the method tree_route_uri:
>
> int tree_route_uri(struct sip_msg * msg, char * _tree, char * _level) {
> 	struct rewrite_data * rd = NULL;
> 	int index;
> 	str ruser;
> 	str ruri;
> 	if (parse_sip_msg_uri(msg) < 0) {
> 		return -1;
> 	}
> 	ruser.s = msg->parsed_uri.user.s;
> 	ruser.len = msg->parsed_uri.user.len;
> 	ruri.s = msg->parsed_uri.user.s;
> 	ruri.len = msg->parsed_uri.user.len;
> 	do {
> 		rd = get_data();
> 	} while (rd == NULL);
> 	index = rd->default_carrier_index;
> 	release_data(rd);
> 	return carrier_rewrite_msg(index, (int)_level, &ruri, msg, &ruser,
> shs_call_id, alg_crc32);
> }
> [..]

Hello Carsten,

yes, you're right. I missed this one in my tests. 
I'll commit a fix later, and thank you for your patch.

> Is it ok, that release_data(rd) is not called at every return statement?
> (e.g. here: LM_ERR("desired routing tree with id %i doesn't exist\n",
> (int)_tree);).

You're here right, too. Fixed in revision 2829.

> If i understood the source right, i will add a patch for this until end
> of the week and add pseudo-variables-support to the
> "cr_tree_rewrite_uri"-Method.

Perhaps you can use the "new" method of evaluating of pseudo-variables, the 
pv_parse and pv_print functions, like in textops:append_to_reply_f 
and :it_list_fixup.

Cheers,

Henning