[Bayonne-devel] Some thoughts on Bayonne drivers and other things

[David Sugar]

Now that I have some time to look at Bayonne 2 drivers again, as the the 
0.6.x releases are nearly complete, I thought I would touch upon several 
different issues where considerable discussion has recently occurred. 
The first of these involve driver features in development such as fax, 
peer signaling for gateways, and continues audio processing; and the 
second involves expanding what drivers will be supported in Bayonne 2 in 
the future.

First, I would like to introduce the feature set for drivers which will 
support continual audio processing.  This allows for audio to be 
overlayed from multiple sources, including the possibility of continues 
music on hold or streaming audio both into and out of Bayonne realtime 
over the internet such as for live audio feeds for internet radio from 
"telephone" callers.  CAP also allows for the creation of continues 
audio logging and the ability to have plugins in the future for audio 
processing, such as soft pam detection.  CAP probably will also be 
needed to support features like MRCP attached TTS/ASR services.

Not all drivers will be capable of continues audio.  The SIP and H.323 
drivers can be made to use of it today.  The existing Voicetronix driver 
certainly can be made to do so as well.  Some drivers that may never do 
continues audio include most CAPI cards, which also suffer from latency 
issues, and voice modems.  I think Peter, in particular, would like to 
see continues audio used in all the new drivers where it can be.

Which brings me to the second topic, new drivers.  In Bayonne I can 
construct drivers that sit above the vendors SDK and supplied user mode 
libraries.  Bayonne 2 now runs successfully on a number of different 
platforms as well, including Microsoft Windows, and various BSD systems. 
 This, combined with the greatly simplified driver model in Bayonne 2, 
makes it much easier to introduce and support Bayonne drivers for many 
more cards (and protocol stacks) than I have in the past.

Some drivers I considered strategic, and hence were developed as 
essential.  SIP certainly was considered necessary to demonstrate voip 
functionality, and the Voicetronix card had the easiest sdk to complete 
and demonstrate basic analog functionality under the new server.  Some 
additional drivers were simply migrated from the 1.x server (H.323, 
CAPI, and Dialogic) but are only worked on as time allows.

One additional driver family I am thinking of is adding zaptel driver 
support.  This particularly is useful not just for using Digium 
hardware, but also because Sangoma is adapting their cards for zaptel, 
and they happen to make their cards available with drivers similarly 
cross-platform, and on at least most of the same platforms 2 Bayonne 
supports today.  Another one is the interesting family of Synway cards 
from China, which are also multi-platform.  Other possibilities that can 
be developed and used in Bayonne, or offered through libbayonne to other 
projects, include Aculab, voice modems, Eicon, Pika, Brooktrout, Natural 
Microsystems, Rhetorix, etc.  Other protocol stacks that are possible 
include iax and ucn.  Again, having the ability to do user level library 
integration makes any of these quite possible to test and implement, 
however, given limited time that is available, I only view a few of 
these strategic enough to pursue unfunded.

Which cards may be tackled and which drivers completed and fully tested, 
as well as how rapidly new driver features are incorporated, will depend 
largely on where time and/or funding are available through the end of 
the year.  The aculab drivers may be the first to be funded, and if so 
they will get a lot of time and attention over the next few months.  As 
there are people in the Chinese development community who are interested 
in the synway cards, at least initial development will be started on 
those, and completed as time allows. Other drivers will entirely depend 
on what interest and/or funding is available to complete.

I hope this covers everything that is current and important with Bayonne 
2 driver development.  I will soon write about the future of Bayonne 2 
application development for 0.7 and beyond.  I will also write soon on 
some ideas for introducing further commercial support for existing 
Bayonne 1.x and future 2 production deployments, as well as for 
potential Bayonne based products and services.  Some of this I may also 
talk about at ClueCon in Chicago next month as well.

David

dyfet.vcf
Description: Vcard