Anthony, thanks for further information!

My goal is not to wait for the break to be executed, I just want to make sure it will be handled properly – and actually break the channel that is playing. So nothing wrong with the async behaviour itself - I like it the way it is.

I’ve been digging into the FS code for a couple of hours, and I think the problem occurs when I’m “fast” enough to add the break command, and then a new playback command in the event queue (and they both exist there). Then in the main loop of switch_ivr_play_file(), after CF_BREAK flag is checked, switch_ivr_parse_all_events() will be called. If both events exist in the queue, the CF_BREAK will first be set, then on the next event it will be reset again, and the new playback will begin. But it will never break the original playback.

Since I havn’t digging too deep in the code, I’m not 100% sure of this :) I was hoping you could help out with your expertice. My current patch looks something like this – do you think it’s possible this causes other problems? By the way – sorry for getting this technical in the users-list, I guess this belongs more to the dev-list...

Index: src/switch_ivr.c

===================================================================

--- src/switch_ivr.c (revision 17000)

+++ src/switch_ivr.c (working copy)

@@ -660,12 +660,16 @@

SWITCH_DECLARE(switch_status_t) switch_ivr_parse_all_events(switch_core_session_t *session)

{

int x = 0;

+ switch_channel_t *channel = switch_core_session_get_channel(session);

switch_ivr_parse_all_messages(session);

- while (switch_ivr_parse_next_event(session) == SWITCH_STATUS_SUCCESS)

+ while (switch_ivr_parse_next_event(session) == SWITCH_STATUS_SUCCESS) {

x++;

+ if (switch_channel_test_flag(channel, CF_BREAK)) {

+ break;

+ }

if (x) {

switch_ivr_sleep(session, 0, SWITCH_TRUE, NULL);

/Peter Olsson

Från: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] För Anthony Minessale
Skickat: den 16 mars 2010 14:55
Till: freeswitch-users@lists.freeswitch.org
Ämne: Re: [Freeswitch-users] ESL and application "break" (to stop file playback) - timing issues.

These events sent via sendmsg are designed to be asynchronous.
artificial waiting may be adding to the problem.

Step outside of what you are trying to do and consider that the sendmsg command itself
is designed to send a 1 way message to the channel, the reply is only to confirm that it was sent not the reply from
the application you are executing, because executing an app is not the only thing you can do.

if you don't like async behaviour you could always send a blocking instruction like the read or play_and_get_digits app to collect your info.

If anything, it might be ok to put this in the "break" app at the end

switch_channel_wait_for_flag(channel, CF_BROADCAST, SWITCH_FALSE, 2000, NULL);

On Tue, Mar 16, 2010 at 6:41 AM, Peter Olsson <peter.olsson@visionutveckling.se> wrote:

I agree, that is a possibility. However, when sending a “SendMsg” with application “break”, and receive “+OK” as it’s reply, you kind of expect that it was received correctly, and will be handled in the event queue properly.

I have a created a small patch that (probably) will fix this in FS, without breaking anything else. I just need to test it first – if everything seems ok I will add a jira for it, with the patch attached.

/Peter

Från: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] För Anthony Minessale
Skickat: den 15 mars 2010 17:21

Till: freeswitch-users@lists.freeswitch.org
Ämne: Re: [Freeswitch-users] ESL and application "break" (to stop file playback) - timing issues.

you could also wait for the execute_complete of the file you were playing

On Mon, Mar 15, 2010 at 9:40 AM, Peter Olsson <peter.olsson@visionutveckling.se> wrote:

Thanks for the reply.

Yes, I thought of this.

But that still shouldn’t solve the issue? It will only tell that the CF_BREAK flag was set on the channel, not that the current playback detected the flag in the loop (which I guess is done in a different thread), or is this event sent once the playback really did stop?

/Peter

Från: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] För Anthony Minessale
Skickat: den 15 mars 2010 16:31
Till: freeswitch-users@lists.freeswitch.org
Ämne: Re: [Freeswitch-users] ESL and application "break" (to stop file playback) - timing issues.

you could wait for the execute_complete event for your break command.

On Mon, Mar 15, 2010 at 3:20 AM, Peter Olsson <peter.olsson@visionutveckling.se> wrote:

I’m not sure if this belongs in Jira, or if it’s possible to do it another way – so I try the list first :)

I’m using ESL to do some IVR functions. Basically I play some sound files, wait for DTMF, and then do something, like record a message, transfer the call to a mobile phone etc. Quite basic and simple.

I’ve noticed one problem with this, and it’s when I want to stop playback using command “break”. In 99% of the time everything works as expected, but sometimes I get this problem – and the problem is that the current file doesn’t stop playing. And I think I know why this is happening.

Lets say I first send playback(tone_stream://%(1500, 3500, 440.0, 0.0);loops=100) to the channel. This will play a ring-tone (swedish) and loop through it 100 times.. When I want to stop this tone I send the break command, and directly after this a new playback command. Maybe 1 time in 100 tries the tone_stream doesn’t stop playing. I think this is becuase how break works. From what I understand from the source it just sets the CF_BREAK flag on the channel, and then the playback application will detect this and stop the playback. However, when I send a new playback command immediately after break it will reset the CF_BREAK-flag again, to make sure it won’t cause any problems for the new playback. So if the original playback (tone_stream) didn’t check the flag before it was reset, it won’t known that it was supposed to stop playing.

I guess I could wait 50ms before sending the new playback message, but I can really never be sure that it was handled properly.

Is there any other way around this? How about “break <uuid> all”, will it make any difference in this case? Or should this just be treated as a timing bug, and filed to Jira? Any help on this would be greatly appreciated.

These are the messages I’m sending;

SendMsg <uuid>

call-command: execute

execute-app-name: playback

execute-app-arg: tone_stream://%(1500, 3500, 440.0, 0.0);loops=100

When I don’t want this to play anymore I send a break;

SendMsg <uuid>

call-command: execute

execute-app-name: break

And immediately after this I want to play another file;

SendMsg <uuid>

call-command: execute

execute-app-name: playback

execute-app-arg: file/to/play.wav

Regards,

Peter Olsson

_______________________________________________
FreeSWITCH-users mailing list
FreeSWITCH-users@lists.freeswitch.org
http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
http://www.freeswitch.org

--
Anthony Minessale II

FreeSWITCH http://www.freeswitch.org/
ClueCon http://www.cluecon.com/
Twitter: http://twitter.com/FreeSWITCH_wire

AIM: anthm
MSN:anthony_minessale@hotmail.com
GTALK/JABBER/PAYPAL:anthony.minessale@gmail.com
IRC: irc.freenode.net #freeswitch

FreeSWITCH Developer Conference
sip:888@conference.freeswitch.org
iax:guest@conference.freeswitch.org/888
googletalk:conf+888@conference.freeswitch.org
pstn:+19193869900