[Asterisk-Users] Digium should develop and sell just Dummy card. For timing...

[Chris Albertson]

Steven,

Good comments but remember good enginerring starts with reading
the requirements and desiging to those requirements.
in the case of SIP at least these is an RFC.  What is the
timming requirement on media packets?  How is the stream
synchronized?  I'll read it in the next few days but I'd
bet a beer there is no requirement for precise timming.

Any discusion about PCI cards, RTC timmers and the like is in
a complete vacuum unless you know what exactly it is that the
software is required to do.


--- Steven Critchfield <critch at basesys.com> wrote:
> On Wed, 2003-10-15 at 11:56, Andrew Kohlsmith wrote:
> > > Some one else here has mentioned the quality of software design
> due to
> > > the need for hardware timing. This should be addressed by the
> fact that
> > > many tools are using hardware timing. Mp3 players use the sound
> device
> > > as a timing source. They can only be feed so much data at a time
> as it
> > > is being serviced. When you can feed it more, you do. In the case
> of the
> > > mp3s on voip only systems, the mp3 player no longer is directly
> coupled
> > > to a device that can control speed. The mp3 player is dumping
> data as
> > > quickly as it can, and as asterisk tosses it into the correct
> format and
> > > gets it out on the ethernet wire, it can then service more data.
> In the
> > > case of Digium hardware, or the appropriate dummy drivers, we get
> a
> > > timing source to directly couple to the channels.
> > 
> > I agree -- I think what that particular poster's point was was that
> there 
> > are already sufficiently jitter-free timing sources in the PC and
> that it 
> > seemed assinine to add another for a paltry 1KHz (IIRC).  He has a
> point, 
> > but (again, speculation) Asterisk boxes may or may not have RTCs or
> even 
> > USB ports, and using a processor-specific timer is even worse for 
> > portability.  By locking on to an add-on card that can be added to
> any 
> > system with a PCI bus, Asterisk gains portability.
> 
> My need to flame is over for the moment, semi rational thoughts
> appear
> below.
> 
> Why make additional hardware whose driver needs be ported to other
> systems when you can make similar dummy interfaces? What happens to
> the
> systems that don't have PCI buses. I can think of some older MAC
> hardware that can run linux, but not PCI cards. Not that I would
> bother
> my self with one of them, but the point stands only to say that
> building
> hardware for a timing source that doesn't have another function is
> stupid. 
> 
> BTW, according to /usr/src/linux/Documentation/rtc.txt, I doubt there
> is
> many systems we will encounter that don't have a rtc.   
> 
> > And again, where is the official Digium response?  This is all
> speculation.  
> > I was just putting out a request to end this stupidity and have an
> official 
> > response so everyone can point to it and say "that's why."
> 
> The official response is in the code if you ever take time to read
> it.
> Since Mark is the chief maintainer of CVS, he probably agreed to
> anything that made it's way into the code base.
> 
> > I didn't think it'd end up in this mini flame war.
> 
> Well at least I stayed out of the email formatting war of late. I've
> been ignoring almost any HTML email, and for the most part quit
> dealing
> with this list.
> -- 
> Steven Critchfield  <critch at basesys.com>
> 
> _______________________________________________
> Asterisk-Users mailing list
> Asterisk-Users at lists.digium.com
> http://lists.digium.com/mailman/listinfo/asterisk-users


=====
Chris Albertson
  Home:   310-376-1029  chrisalbertson90278 at yahoo.com
  Cell:   310-990-7550
  Office: 310-336-5189  Christopher.J.Albertson at aero.org
  KG6OMK

__________________________________
Do you Yahoo!?
The New Yahoo! Shopping - with improved product search
http://shopping.yahoo.com