Re: [Linphone-developers] lots of bugs resolution and multicalls improvements

[Simon Morlat]

Aurélien, Maxim,

The current code has bugs in these scenarios. I will work on these just
after the next release. Instead of adding more call states, I plan to
use the stream direction information (recvonly, sendonly, sendrecv) to
decide what to do in each case.
Maybe just one state LinphoneCallInactive would be necessary to add.

Simon

Le mardi 01 février 2011 à 12:54 +0000, Aurelien Bouin a écrit :
> Hi,
> 
> 
> The first email I sent about this subject was attached with a patch
> that potentially solve the problem... But I agree that it leads to big
> complexity so now I am waiting for a better implementation ... because
> I don't have much time to work on that since it is working for me ...
> Regards,
> 
> 
> Aurélien
> 
> --- En date de : Mar 1.2.11, Maxim Podbereznyy <address@hidden> a
> écrit :
>         
>         De: Maxim Podbereznyy <address@hidden>
>         Objet: Re: [Linphone-developers] lots of bugs resolution and
>         multicalls improvements
>         À: address@hidden
>         Date: Mardi 1 février 2011, 12h23
>         
>         Hi Simon, Aurelien!
>         
>         I agree with both of you! :)
>         
>         Is this behavior already in git or requires any commits?
>         Please clarify
>         
>         best regards,
>         Max
>         
>         2011/2/1 Aurelien Bouin <address@hidden>
>                 Hi Simon,
>                 I agree that the correct behaviour would be as
>                 fallows:
>                 
>                 A pauses B (thus B listens to music)
>                 B pauses A (thus the call is totally inactive)
>                 B resumes A (this B listens to music again)
>                 A resumes B (the call is now full duplex).
>                 
>                 
>                 and it also could be :
>                 
>                 A pauses B (thus B listens to music)
>                 B pauses A (thus the call is totally inactive)
>                 A resumes B (this A listens to music again)
>                 B resumes A (the call is now full duplex).
>                 
>                 
>                 Regards,
>                 
>                 Aurelien
>                 
>                 --- En date de : Mar 1.2.11, Simon Morlat
>                 <address@hidden> a écrit :
>                 
>                 > De: Simon Morlat <address@hidden>
>                 > Objet: Re: [Linphone-developers] lots of bugs
>                 resolution and multicalls improvements
>                 > À: address@hidden
>                 
>                 > Date: Mardi 1 février 2011, 10h54
>                 
>                 > Hi Aurélien, Maxim,
>                 >
>                 > I agree that the correct behaviour would be as
>                 follows:
>                 >
>                 > A pauses B (thus B listens to music)
>                 > B pauses A (thus the call is totally inactive)
>                 > B resumes A (this B listens to music again)
>                 > A resumes B (the call is now full duplex).
>                 >
>                 > It would be a bad thing if B resumes A changes the
>                 call in
>                 > full duplex.
>                 > A must always decide whether it can send and listen
>                 real
>                 > audio.
>                 >
>                 > Everybody agrees with that ?
>                 >
>                 > Simon
>                 >
>                 > Le mercredi 26 janvier 2011 à 07:56 +0000, Aurelien
>                 Bouin
>                 > a écrit :
>                 > > Hi Simon
>                 > >
>                 > > > De: Simon Morlat <address@hidden>
>                 > > > Objet: Re: [Linphone-developers] lots of bugs
>                 > resolution and multicalls improvements
>                 > > > À: address@hidden
>                 > > > Cc: "Aurelien Bouin" <address@hidden>
>                 > > > Date: Mardi 25 janvier 2011, 21h35
>                 > > > Hi Aurélien,
>                 > > >
>                 > > > Thanks for the patches, it took some time to
>                 > review them.
>                 > > Yeah I guess it's quite a lot of work while you
>                 are
>                 > probably very busy ... sorry for that . . .
>                 > >
>                 > > > I merged the one concerning mediastreamer2,
>                 > however I feel
>                 > > > more
>                 > > > unconvinced by the one in liblinphone,
>                 especially
>                 > with the
>                 > > > multiple new
>                 > > > LinphoneCallState that it introduces.
>                 > > >
>                 > > > It leads to big complexity, I hope there is a
>                 > more simpler
>                 > > > way to solve
>                 > > > the problem you were attempting to solve. Could
>                 > you
>                 > > > describe a
>                 > > > bit more the actual problems you were facing ?
>                 > > Yes you are right it's really too complicate too
>                 > handle multicalls with this kind of
>                 implementation...
>                 > > The problem is easy to be reproduced between two
>                 > Linphone... :
>                 > > Call each other
>                 > > The streaming should be running
>                 > > Execute a pause in one side
>                 > > Execute a pause on the other side
>                 > > Here it should be OK each one is in Pause status
>                 > > But when one ff them Resume the call the call will
>                 > restart while the other is still in Pause...which is
>                 a
>                 > problem... The new states help to handle those
>                 cases...but
>                 > it's really complex to implement...
>                 > > Give it a try whenever you have the time, if you
>                 have
>                 > a better implementation I would appreciate it...
>                 > >
>                 > > >
>                 > > > Concerning the ability to start several calls on
>                 > hold, I'm
>                 > > > ok with that.
>                 > > > However the terminology in liblinphone is
>                 > "paused" to refer
>                 > > > to putting a
>                 > > > call on hold, so I would suggest the new
>                 > parameter in
>                 > > > LinphoneCallParams
>                 > > > would use the same terminology.
>                 > > > Also, I think there is no reason to modify the
>                 > sal_call()
>                 > > > prototype,
>                 > > > since the nature of the call (paused or not) is
>                 > only the
>                 > > > SalMediaDescription describing the offer, that
>                 > would
>                 > > > contain send-only
>                 > > > streams instead of sendrecv ones.
>                 > > No problem I will change hold by pause when I will
>                 > have the time though...
>                 > >
>                 > >
>                 > > >
>                 > > > Best regards,
>                 > > >
>                 > > > Simon
>                 > > >
>                 > > Thank you Simon for your precious time, do not
>                 > hesitate to ask anything ...
>                 > > Regards,
>                 > >
>                 > > Aurélien BOUIN
>                 > >
>                 > > >
>                 > > >
>                 > > > Le mercredi 19 janvier 2011 à 10:11 +0000,
>                 > Aurelien Bouin
>                 > > > a écrit :
>                 > > > > Hi all,
>                 > > > > I have not been able to work on linphone for
>                 > such a
>                 > > > lot of time, but we have seen some problems
>                 since
>                 > multiple
>                 > > > calls are implemented...
>                 > > > > Here is a non exhausting list of features
>                 > implemented
>                 > > > in the attached patchs :
>                 > > > > ->resolve problem of bipartite hold
>                 > on/off, some
>                 > > > states have been added, but makes the callbacks
>                 > hard to be
>                 > > > understood ...
>                 > > > > ->instability into thread joined into
>                 > msv4l2 and
>                 > > > msthread
>                 > > > > ->being able to do multiples calls with
>                 > the
>                 > > > possibility of doing them hold on...
>                 > > > > ->being able to receive DTMF from any
>                 > call
>                 > > > process...
>                 > > > > ->sal_media_description_has_dir seems to
>                 > not being
>                 > > > functional like it was
>                 > > > > ->improvements with the plugins use under
>                 > windows
>                 > > > >
>                 > > > > I hope that you will have the time to
>                 > consider those
>                 > > > changes, do not hesitate to ask ...
>                 > > > > Thank you for your time spent to work on
>                 > that ...
>                 > > > >
>                 > > > > Aurélien BOUIN
>                 > > > >
>                 > > > >
>                 > > > >       
>                 > > > >
>                 > _______________________________________________
>                 > > > Linphone-developers mailing list
>                 address@hidden
>                 > > >
>                 http://lists.nongnu.org/mailman/listinfo/linphone-developers
>                 > > >
>                 > > >
>                 > > >
>                 > >
>                 > >
>                 > >       
>                 > >
>                 > > _______________________________________________
>                 > > Linphone-developers mailing list
>                 > > address@hidden
>                 > >
>                 http://lists.nongnu.org/mailman/listinfo/linphone-developers
>                 >
>                 >
>                 >
>                 > _______________________________________________
>                 > Linphone-developers mailing list
>                 > address@hidden
>                 >
>                 http://lists.nongnu.org/mailman/listinfo/linphone-developers
>                 >
>                 
>                 
>                 
>                 
>                 _______________________________________________
>                 Linphone-developers mailing list
>                 address@hidden
>                 http://lists.nongnu.org/mailman/listinfo/linphone-developers
>                 
>         
>         
>         
>         -----La pièce jointe associée suit-----
>         
>         _______________________________________________
>         Linphone-developers mailing list
>         address@hidden
>         http://lists.nongnu.org/mailman/listinfo/linphone-developers
>         
> 
> _______________________________________________
> Linphone-developers mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/linphone-developers