[asterisk-dev] Another dial-option, catching hangup of caller party

[Johan Wilfer]

Could someone who knows app_dial.c better than I look at this?
It's a very small patch... See bug #0011954

> >>> I wonder will this work with queues. I suspect that queue will try to
> >>> terminate created channel, or at least update some log.
> > I also meant the local channels that issue Dial. However i think,
> > there's the difference, because call is bridged trough queue, so queue
> > would first get notification when caller hanged up. If you intend to
> > use it in queue, can you give us an update when you'll have it
> > working?
> >   
> When dial() is called directly outside of queue, it works.
> In the queue it only works if /n is appended to the local channel, 
> otherwise it doesn't.
> 
> Why this is so and what it means I don't really understand. When I use 
> the G-flag I don't need to
> specify /n after the local channel.
> 
> 
> I've done some tests and I'm very satisfied with the results. I've 
> encounterd one strange side effect however:
> I got this message when using the new flag:
> [Feb 11 16:30:44] NOTICE[22370]: app_dial.c:1845 dial_exec_full: PEER 
> context: incoming; PEER exten: ;  PEER priority: 1
> 
> The line's in question are these:
> --------------------------------------------------------------
> 1844:                if (ast_test_flag64(&opts, OPT_PEER_H)) {
> 1845:                        ast_log(LOG_NOTICE, "PEER context: %s; PEER 
> exten: %s;  PEER priority: %d\n",
> 1846:                                peer->context, peer->exten, 
> peer->priority);
> 1847:                }
> --------------------------------------------------------------
> 
>  From my patch:
> --------------------------------------------------------------
> @@ -262,12 +264,13 @@
>      OPT_CALLEE_GOSUB =      (1 << 28),
>      OPT_CALLEE_MIXMONITOR = (1 << 29),
>      OPT_CALLER_MIXMONITOR = (1 << 30),
> +    OPT_CALLEE_GO_ON =      (1 << 31),
>  };
>  
> -#define DIAL_STILLGOING            (1 << 31)
> -#define DIAL_NOFORWARDHTML        ((uint64_t)1 << 32) /* flags are now 
> 64 bits, so keep it up! */
> -#define OPT_CANCEL_ELSEWHERE    ((uint64_t)1 << 33)
> -#define OPT_PEER_H    ((uint64_t)1 << 34)
> +#define DIAL_STILLGOING        ((uint64_t)1 << 32)
> +#define DIAL_NOFORWARDHTML        ((uint64_t)1 << 33) /* flags are now 
> 64 bits, so keep it up! */
> +#define OPT_CANCEL_ELSEWHERE    ((uint64_t)1 << 34)
> +#define OPT_PEER_H               ((uint64_t)1 << 35)
>  
>  enum {
>      OPT_ARG_ANNOUNCE = 0,
> --------------------------------------------------------------
> 
> Right now my patch has to be applied manually because a whitespace 
> commit since yesterday.

And this one?

> Can someone explain why almost every flags are defined inside an enum 
> but some are #define:
> DIAL_STILLGOING, DIAL_NOFORWARDHTML, OPT_CANCEL_ELSEWHERE, OPT_PEER_H

/Johan