[asterisk-dev] res_pjsip: sending to arbitrary URI

Kevin Harwell kharwell at digium.com
Fri Nov 22 11:45:01 CST 2013


On Tue, 2013-11-12 at 11:38 -0600, Kevin Harwell wrote:
> Hello,
> 
> So I think this was mentioned at AstriDevCon 2013 - Do we want to or
> should we be able to send outbound SIP messages to an arbitrary URI vs
> always having to specify an endpoint?
> 
> Currently in res_pjsip land in order to send stuff outbound an endpoint
> is required.  Since there will be cases where one may need to send to a
> URI with no associated endpoint this poses a problem.
> 
> One solution is to just implement the ability to send to an arbitrary
> URI.  Unfortunately, this doesn't exactly fit in with the current way of
> doing things and might lead to some unexpected side effects.  Also this
> might lead people wondering why this functionality is allowed in one
> place and not another (inconsistency not good).
> 
> Two other solutions have been suggested by Mark Michelson that fit in
> better with the current framework:
> 
> 1) Use a specially-named endpoint (maybe called "default_outbound").
> This endpoint can be automatically created when res_pjsip is loaded and
> contain nothing but the default values for the endpoint. If people want
> to tweak default behavior, then they can create an endpoint called
> "default_outbound" in their pjsip.conf file and set appropriate values
> on it. This approach has the advantage of "just working" out of the box
> and allowing for overriding of the default behavior if desired.
> 2) Create a new option for PJSIP global configuration (maybe called
> "default_outbound_endpoint") that indicates an endpoint to be used when
> sending an outbound request to a URI. This approach has the advantage of
> not creating any "secret" endpoints that the user did not explicitly
> place in the configuration file.
> 
> I kind of like #2 as there is nothing hidden and it allows the end user
> to name the default outbound endpoint.  Thoughts? Other ideas?
> 
> Thanks,
> 

I am planning on moving forward with #2 soonish unless I hear of some
major reason why not to.




More information about the asterisk-dev mailing list