Hello,

On 12/28/06, yuhang <[EMAIL PROTECTED]> wrote:
> I would like to know back ground of sipXtapi: the history of sipXtapi
> and  the history of Libs it relies on, successful commercial products
> based on  sipXtapi, and stories (if any) of people who dedicate to sipXtapi.
I'm rather new to this project and could not give you comprehensive
answer. Old Pingtel emploies may be asked for more.. As it was said here:
http://sipx-wiki.calivia.com/index.php/SipX_-_Open_Source_PBX#History

"Development of sipX started in 1999 and was first introduced as a
commercial product by Pingtel Corp. Early 2004 Pingtel adopted an
open-source business model, became a founding member of SIPfoundry and
made available the entire commercial code base as an open source
project under the name of sipX. Since then an active community has
formed around sipX with Pingtel doing all its development efforts as
part of the open source sipX project."

I belive that all sipX* libraries was originally developed in Pingtel before
they become Open-Source.

The biggest project, which use sipXtapi is AOL Triton. But they use sipXtapi
with GIPS libraries, not with open-source sipXmediaLib. And I could say
that sipXmediaLib is the least mature library between others. There are
many thing to be done in it before it could really compete with commercial
libraries like GIPS. But this open big and fun field for application
of our skills.
That's not so bad, right? ;)


> below is 2 problems i run into when playing with sipXtapi.
>
> 1) 100% cpu usage after pluging out RJ45
> I call sipxInitialize and    then plug out RJ45 so that network is
> no longer available. Seconds later, cpu    is 100% full and i pause
> the app and see:
<Skipped>
> Is this a new problem? Is    there any solutions?
This problem was already reported:
http://track.sipfoundry.org/browse/XMR-79

I still have no time to dig into this problem. If somebody will write
appropriate patch, it will be very appreciated.

> cutting sound
> I establish call with sipXezPhone, make cpu    high-load with other
> application. Sure, the sound from sipXezPhone is cutting    at this
> moment. Then, i stop burning the cpu, but sometimes sipXezPhone
> can    not immediately recover to play fluent sound (the sound is cutting).
> Sometimes, the cutting sound even goes on for several minutes. Codec
> is 711.
> Given the fact that the library was originally created for use in    hardware
> phone, is the timing mechanism robust enough for Win32 environment
> where other threads may heavily use cpu?
Yes, timing is not very precise under Windows. This cause very bad audio
quality in some cases. :( For comparison, sound quality under Linux is much
better. In my vision this problem include two major parts: precise timing and
audio I/O code. I propose Multimedia Timers for at least for internal
sipXmediaLib usage under Windows to solve precise timing part of the
problem. Audio I/O part have two (still unfinished) solutions - increase
priority of I/O threads (as described in XMR-70 issue)  or implement
totally different audio I/O. Patch attached to XMR-70 issue need hard review,
I'm not sure that all thing done correctly. Once I have free time I'll
review it.
DirectSound I/O was implemented by Charlie Hedlin for his sipXtapiDotNet
project, but still not ported back to sipXtapi. I'm waiting his patch.

And one more point where audio could get choppy is MprDejitter/MprDecoder/
MprJitterBuffer. One issue with them was reported as XMR-80, but in high
load case this is not snough, I think. But this is place for research - I never
did such tests. You're welcome to begini this work. ;)

-- 
Regards,
Alexander Chemeris.
_______________________________________________
sipxtapi-dev mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipxtapi-dev/

Reply via email to