Re: [Linphone-developers] linphone blackberry error 523 on device

[Antonio Grillo]

Hi Jean,

unluckily I've not solved the problem. I'm trying to tune the
parameters of the read() method in RecvStream.java in order to vary
the buffering, minimum data returned policies without great results.

Just in one trial I reached 1 minute and 2 second of voice between two
blackberry devices.

My solution, as aleksei suggested is not a real solution follows:

create this class as inner class of AudioStreamImpl.java

class RecvStreamTimer extends TimerTask {
                public void run() {
                        mRecvStream.stop();
                        mRecvStream=new RecvStream(mSession,mSendStream);
                        Thread rTmpThread = new Thread(new Runnable(){
                                public void run() {
                                        mRecvStream.start();                    
        
                                }
                        });
                        rTmpThread.start();
                }
}

Add the following lines after the Timer that print the statistics:

Timer recvTimer = new Timer();
recvTimer.scheduleAtFixedRate(new RecvStreamTimer(), 25000, 25000);

in this way you have some second of silence for each restart.

Aleksei says to me that he have successfully realized a voip call
between a BlackBerry and an HTC device using a transcoding SIP server.
This is very strange, the problems occurs just when considering two
blackberry devices.

Hope this helps!!

Antonio



On Tue, May 10, 2011 at 19:31, Jehan Monnier <address@hidden> wrote:
> Hi Antionio,
> Very interesting indeed.
> Just a question, at the end, do you manage to fix the issue ?
> It would be very kind if you can post a patch that I can integrate to the
> main GIT tree.
> Cheers
> Jehan
> www.linphone.org
>
>
>
>
>
> Le 10 mai 2011 à 10:59, Antonio Grillo a écrit :
>
> Dear Aleksei,
>
> I'm trying to investigate the problem in the RecvStream.java class.
>
> I have fixed the playerupdate() method, the states of the player are
> represented as Strings and so they have to be compared using equals()
> instead of ==.
>
> Once fixed I have obtained that player at 30 secs of normal call
> enters in the PlayerListener.ERROR state with EventData code -5 that
> is an unspecified error.
> (http://supportforums.blackberry.com/t5/Java-Development/PlayerListener-ERROR-eventData-is-5-But-PlayerListener-ERROR/td-p/464383)
>
> As posted by Izrdus in the link above, he has idenitifed the player
> error problem as follows:
>
> "I had a problem with my bookkeeping when the offset approaches length
> in buffer javax.microedition.media.protocol.SourceStream.read(byte[],
> int, int). Basically after 58000 bytes the audio would become
> corrupted, causing the player to raise an error."
>
> What do you about the Izrdus' solution?
>
> I found an official code snippet from BlackBerry
> (http://www.blackberry.com/knowledgecenterpublic/livelink.exe/fetch/2000/348583/800332/1089414/How_To_-_Support_streaming_audio_to_the_media_application.html?nodeid=1369914&vernum=0)
> explaining how to use the SourceStream as source for a Player. In that
> snippet the Thread.sleep() is used with a delay of 100ms so I think
> that sleeping 20ms for collecting more data should not be the problem.
>
> Cheers
>
> Antonio Grillo, PhD Student
> Dept. of Computer Science, System and Industrial Engineering
> Room D1-21(aka S4ALab)
> phone: 06 72597714
> mail: address@hidden
> University of Rome "Tor Vergata"
> Via del Politecnico, 1 00133 Roma
>
>
>
> On Fri, Apr 29, 2011 at 04:33, aleksei drassen <address@hidden> wrote:
>
> The problem is in the player buffer (RecvStream.java), when the player is
>
> ready to "play"  the buffer (byte[] b) of the player does'nt have anything
>
> data stored yet, then the outputstream returns -1 (end of stream) throw to
>
> the thread sleep(20), so the player enter to a error state and stops, i'm
>
> trying to determine how i can solve this without affect other app layers,
>
> one solution (stable) is putting the recvstream in a timer to reset
>
> and initiate the class ever 30 secs, but this is a bad solution (is'nt a
>
> solution).
>
> so, i think the solution maybe is rewritting the class take care to the
>
> delayer timer and write stream implementation. For now i have some work in
>
> my house and i cannot write anything for the moment, but if you understand
>
> what is i say i think you can write the best stable class implementation
>
> (and i appreciate if you share anyway)
>
>
> lucky you!
>
> From: address@hidden
>
> Date: Thu, 28 Apr 2011 13:59:26 +0200
>
> Subject: Re: [Linphone-developers] linphone blackberry error 523 on device
>
> To: address@hidden
>
> Hi Aleksei,
>
> I'm trying linphone-bb on two real devices (Bold 9700).
>
> I have patched the source code and I've disabled the log (ALT +
>
> twice("lg")) as you suggested in the last mail.
>
> Now, I can make VoIP call between two real devices but at 30 sec
>
> (sometime a little more) there is no more audio, I'm trying to
>
> understand the classes SendStream.java and RecvStream.java.
>
> Do you make some progress in solving this problem?
>
> If you have detected a possible source of problem may I help you in
>
> investigate and solve it?
>
> Cheers
>
> Antonio
>
>
> On Fri, Apr 8, 2011 at 18:03, aleksei drassen <address@hidden>
>
> wrote:
>
> Hi, no, i test all and is no rigth to prevent the 523 error on the
>
> device,
>
> the error ocurrs when the device try to fill queue (thread(20))  ms on
>
> the device log, so, you'll need go directly to the device and disable
>
> (on
>
> hand) the debug log.
>
>
> device: on the home screen press alt+lglg  (press "l" + "g" twice)  and
>
> toggle hide all elements on the list and select global warning, save,
>
> and
>
> thats all.
>
> Maybe you have no sound at 30 secs of normal call (blackberry to
>
> blackberry), but i don't resolve that issue yet.
>
>
> cheers
>
> Date: Thu, 7 Apr 2011 16:30:16 +0200
>
> From: address@hidden
>
> To: address@hidden; address@hidden
>
> Subject: RE: [Linphone-developers] linphone blackberry error 523 on
>
> device
>
> Hi Aleksei,
>
> I tried to implement the patch that you suggested for avoiding the 523
>
> App Error on blackberry.
>
> Unfortunately such patch does not work for me:
>
> 1) I've encapsulated the send method in a thread [Class
>
> AudioStreamImpl.java]
>
> Thread l1TmpThread = new Thread(new Runnable(){
>
> public void run() {
>
> mSendStream=new SendStream(mSession);
>
> mSendStream.start();
>
> }
>
> });
>
> l1TmpThread.start();
>
> 2) I've set a 30 ms sleep time when an RTPException occurs [Class
>
> SendStream.java]
>
> try {
>
> mSession.sendPacket(packet, mTs);
>
> } catch (RtpException e) {
>
> Thread.sleep(30);
>
> sLogger.disableLogLevel(1);
>
> }
>
> 3) I've disabled all the log features [Class Logger.java]
>
> public void log(int level, String msg, Throwable e){
>
> //No Operations
>
> }
>
> maybe I'm forgotting something??
>
> Thanks in advance for your attentions
>
> Antonio
>
>
>
> _______________________________________________
> Linphone-developers mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/linphone-developers
>
>
> _______________________________________________
> Linphone-developers mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/linphone-developers
>
>